Ensemble attacks integrate the outputs of surrogate models with diverse architectures, which can be combined with various gradient-based attacks to improve adversarial transferability. However, previous work shows unsatisfactory attack performance when transferring across heterogeneous model architectures. The main reason is that the gradient update directions of heterogeneous surrogate models differ widely, making it hard to reduce the gradient variance of ensemble models while making the best of individual model. To tackle this challenge, we design a novel ensemble attack, NAMEA, which for the first time integrates the gradients from the non-attention areas of ensemble models into the iterative gradient optimization process. Our design is inspired by the observation that the attention areas of heterogeneous models vary sharply, thus the non-attention areas of ViTs are likely to be the focus of CNNs and vice versa. Therefore, we merge the gradients respectively from the attention and non-attention areas of ensemble models so as to fuse the transfer information of CNNs and ViTs. Specifically, we pioneer a new way of decoupling the gradients of non-attention areas from those of attention areas, while merging gradients by meta-learning. Empirical evaluations on ImageNet dataset indicate that NAMEA outperforms AdaEA and SMER, the state-of-the-art ensemble attacks by an average of 15.0% and 9.6%, respectively. This work is the first attempt to explore the power of ensemble non-attention in boosting cross-architecture transferability, providing new insights into launching ensemble attacks.

Serving ML models with serverless computing has become increasingly popular in recent years. Many of today's cloud vendors have provided GPU functions to meet the performance requirements of different ML scenarios. However, existing production FaaS platforms suffer from GPU under-utilization and high cloud costs due to poor GPU resource management. In this paper, we propose gShare, an on-demand and efficient GPU function management policy in FaaS platforms. gShare provides a fine-grained GPU virtualization solution for a VM-based multi-tenant FaaS environment. It further decouples the GPU resource from the existing CPU-oriented function management paradigm, enabling flexible GPU sharing across tenants. With a user-transparent vGPU remapping design and aggressive request scheduling policy, gShare can significantly improve the cost-efficiency of GPU functions without causing appreciable function performance degradation. Experimental results show that gShare can reduce GPU usage by 43%–63% compared to the baseline while meeting more than 95% of user latency targets. Compared with the state-of-the-art method, it can also reduce cloud costs by 24%-58% while maintaining better function performance, benefiting both the cloud provider and users.

Video diffusion models (VDMs) have demonstrated remarkable capabilities in text-to-video (T2V) generation. Despite their success, VDMs still suffer from degraded image quality and flickering artifacts. To address these issues, some approaches have introduced preference learning to exploit human feedback to enhance the video generation. However, these methods primarily adopt the routine in the image domain without an in-depth investigation into video-specific preference optimization. In this paper, we reexamine the design of the video preference learning from two key aspects: feedback source and feedback tuning methodology, and present OnlineVPO, a more efficient preference learning framework tailored specifically for VDMs. On the feedback source, we found that the image-level reward model commonly used in existing methods fails to provide a human-aligned video preference signal due to the modality gap. In contrast, video quality assessment (VQA) models show superior alignment with human perception of video quality. Building on this insight, we propose leveraging VQA models as a proxy of humans to provide more modality-aligned feedback for VDMs. Regarding the preference tuning methodology, we introduce an online DPO algorithm tailored for VDMs. It not only enjoys the benefits of superior scalability in optimizing videos with higher resolution and longer duration compared with the existing method, but also mitigates the insufficient optimization issue caused by off-policy learning via online preference generation and curriculum preference update designs. Extensive experiments on the open-source video-diffusion model demonstrate OnlineVPO as a simple yet effective and, more importantly, scalable preference learning algorithm for video diffusion models.

The explosive growth of multimodal web data demands communication that transmits meaning rather than raw bits. Existing semantic-communication systems often fail under noise, missing modalities, and distribution shifts because they optimize surface features instead of modality-invariant knowledge. We present Grasp, a knowledge-centric framework for cross-modal communication. Grasp segments streams into semantic blocks and builds a graph over them; a lightweight Graph Neural Networks (GNN) produces schedulable, importance-weighted representations. At its core is knowledge purification : we minimize a conditional mutual information upper bound to perform a three-way disentanglement—strongly related, weakly related, and task-irrelevant components—so that only essential semantics are transmitted while non-essential factors are suppressed. To maintain synchrony, we introduce one-to-two temporal contrastive learning to achieve triple alignment of video, audio, and text despite sampling asynchrony. For efficient transmission, Grasp uses a cross-modal shared vector-quantization codebook—a discrete knowledge codebook —updated by multimodal attention. At the receiver, a soft-recovery mechanism leverages this shared knowledge to robustly reconstruct semantics under low signal-to-noise ratio (SNR) or missing modalities, yielding graceful degradation. Across web tasks—including cross-modal retrieval and missing-modality inference—Grasp improves knowledge consistency, semantic fidelity, and downstream performance over strong baselines while maintaining low latency. These results show that communication structured around purified knowledge is key to building robust, semantic-aware systems for the modern web.

Integrated Sensing and Communications (ISAC) systems face challenges in balancing sparse human resource allocation with high-quality data acquisition, especially in scenarios involving highly mobile workers and uneven task coverage. While Sparse Mobile CrowdSensing (SMCS) offers a resource-efficient solution by inferring unobserved data from partial sensing tasks, two critical issues remain unaddressed: (1) existing data inference methods often assume collected data is trustworthy, neglecting the detrimental impact of malicious workers on data quality, and (2) the lack of privacy-preserving mechanisms in sparse ISAC may deter worker participation, compromising service quality. To address these, we propose a Quality-driven Privacy-preserving Truth Discovery scheme (QPTD), which optimizes sparse sensing and data quality with privacy constraints in ISAC systems. QPTD integrates a trust-aware Deep Reinforcement Learning (DRL) algorithm for dynamic cell selection, prioritizing cells with high-trust workers and minimizing recruitment costs, while a matrix encryption protocol ensures end-to-end privacy protection for both workers and tasks. By leveraging Matrix Factorization (MF) on encrypted sparse data aggregated from high-trust workers, QPTD infers unobserved tasks and safeguards sensitive information. Theoretical analysis proves the robustness of our privacy guarantees against adversarial attacks. Evaluations on real-world datasets demonstrate that QPTD improves ISAC system performance by increasing total data quality by 1.73%–142.07%, reducing estimation errors by 24.82%–75.32%, and cutting recruitment costs by up to 62.03% while ensuring service performance.

Abstract Stochastic bilevel optimization finds widespread applications in machine learning, including meta-learning, hyperparameter optimization, and neural architecture search. To extend stochastic bilevel optimization to distributed data, several decentralized stochastic bilevel optimization algorithms have been developed. However, existing methods often suffer from slow convergence rates and high communication costs in heterogeneous settings, limiting their applicability to real-world tasks. To address these issues, we propose two novel decentralized stochastic bilevel gradient descent algorithms based on simultaneous and alternating update strategies. Our algorithms can achieve faster convergence rates and lower communication costs than existing methods. Importantly, our convergence analyses do not rely on strong assumptions regarding heterogeneity. More importantly, our theoretical analyses clearly disclose how the computation and communication regarding the Hessian-inverse-vector product under the heterogeneous setting affects the convergence rate. To the best of our knowledge, this is the first time such favorable theoretical results have been achieved with mild assumptions in the heterogeneous setting. Furthermore, we demonstrate how to establish the convergence rate for the alternating update strategy when combined with the variance-reduced gradient. Finally, experimental results confirm the efficacy of our algorithms.

Microservices are widely shared in production user-facing applications. These shared microservices have various resource usage patterns when queries from different call graphs of different services access them. However, existing microservice management works fail to efficiently scale resources for them, mainly due to the lack of fine-grained scheduling of diverse queries. We therefore propose Delphinus , a runtime system that efficiently manages resources for shared microservices while ensuring the Quality-of-Service (QoS). Delphinus comprises a group-oriented query scheduler and a borrowing-based load adapter . The query scheduler identifies diverse queries, groups the containers of shared microservices, and schedules the queries into separate groups. The load adapter efficiently scales resources for shared microservices, and fully utilizes the idle containers among groups when the loads of diverse queries change. Results show that Delphinus reduces CPU and memory usage by 40.1% and 36.4% for shared microservices, respectively, compared to state-of-the-art works.

Function caching is one of the fundamental techniques in FaaS platforms to alleviate coldstart overhead. However, as cache instances consume significant cloud resources (e.g., memory), it is challenging to balance function performance and cache cost. Current systems use simple and rudderless cache polices with a “local cache control” design, which ignores function characteristics such as workload skewness from hot functions and results in either cache contentions or cache resource waste.In this paper, inspired by software-defined networks, we proposeFlame, an efficient cache system to manage cached functions with hotspot-aware instance scheduling and cache allocation. It consists of a two-layer design. Firstly, by decoupling the cache control plane from worker nodes and introducing a centralized cache controller,Flamecan schedule functions from a global view of the cluster’s status, thereby reducing inter-node workload skew. Second,Flamedivides the prior monolithic cache pool within each node into multiple partitions and dynamically assigns them to different hot functions, thereby further mitigating intra-node cache contention. Experimental results from realworld workloads show thatFlamecan reduce cache resource usage by 36% on average while improving function performance by nearly 7× compared to the state-of-the-art method.

Transferable adversarial examples (AEs) are visually indistinguishable from benign images, but can successfully mislead unknown deep neural networks. However, existing AEs normally vary considerably from benign images in the feature space, making them hard to pass label checking and adversarial detection. Therefore, how to make AEs camouflaged, disguising as benign images during detection is still an open problem. In this paper, we propose a novel camouflaged adversarial attack (CAA), which produces camouflaged adversarial examples (CAEs) for the first time. Our main idea is to make CAEs’ adversarial properties keep “dormant” state until the target model inadvertently triggers the “activated” state. To this end, we craftattackandcamouflageperturbations, so that CAEs are visually and feature/label-wise indistinguishable from benign images at first, but will implicitly turn into AEs once being triggered. Specifically, we exploit two common preprocessing operations, image scaling and JPEG compression, as the trigger, and propose a two-stage optimization strategy. As the preprocessing details of target models are unknown, the first stage trains a well-designed generative adversarial network under varying scaling/compression parameters to enhance the robustness of attack perturbations. The second stage uses feature (dis)similarities and contrastive distances to improve the transferability of camouflage perturbations. Extensive experiments on ImageNet dataset validate the effectiveness of CAA. Especially for robust models, the average fooling rate after preprocessing could reach 96.3% outperforming the state-of-the-art adversarial attack by 13.5%.

Knowledge Tracing (KT) aims to predict learners' future responses by modeling historical learning sequences. However, existing works often overlook the severe disruptions in temporal and interaction continuity arising from large time intervals between interactions, which leads to significant prediction errors. Therefore, we identify and validate thisbreakpointeffect through empirical studies, and try to address it via data augmentation (DA). Existing DA methods usually emphasize sequence diversity or similarity while neglecting the breakpoint effect in KT. Hence, we strive to enhance knowledge tracing via breakpoint-aware sequence augmentation and propose Collaborative Learning Path Augmentation (CLPA), a systematic three-phase method to enhance models' representational capacity and alleviate the breakpoint effects. First, identifying boundary interaction pairs that exhibit the temporal gap and response inconsistency. Second, leveraging cross-sequence collaborative information to infer functional pseudo-paths between identified boundaries. Third, reconstructing and inserting pseudo-learning paths into target sequences to restore sequence continuity; moreover, two refinements of the sampling quantity constraint and the distance-based sampling strategy are proposed to ensure augmentation quality. Comprehensive experiments on four real-world datasets validate the effectiveness of CLPA, which significantly enhances the target KT model's representational capacity and alleviates the breakpoint effects.

Cloud-fog Federated Learning (FL) is promising for collaborative model training in large-scale edge networks. In cloud-fog FL with constrained communication resources, selecting high-quality client models for aggregation is critical to boost the global model. However, the clients in real world hold multimodal and heterogeneous data, while existing selection strategies rarely consider the imbalance of communication cost and model convergence rates across modalities, thus seriously degrading the efficiency of model aggregation. Moreover, most previous multimodal fusion methods require aligned multimodal samples. However, the data of modality-heterogeneous clients may be isolated and unaligned in FL, so these previous methods cannot be applied to such scenarios, thereby hindering the knowledge fusion across various modalities. To address the above issues, we propose a cloud-fog FL system named COOL, which achievesunimodal aggregationat the fog layer andmultimodal fusionat the cloud layer. First, we propose a Modality-aware Online Client Selection (MOCS) strategy to assist unimodal aggregation. Unlike previous selection strategies, MOCS realizes the dynamic selection budget allocation for various modalities by monitoring the convergence gap of modalities, thus striking the performance balance among modality-heterogeneous clients. Second, to overcome the limitation of previous methods that require aligned multimodal data, we propose a Multimodal Fusion strategy with Feature Synthesis (MFFS). MFFS realizes multimodal fusion with isolated samples via adaptive feature synthesis and cross-modal attention training, thus building a more powerful multimodal predictor while preserving the client data privacy. Finally, experimental results demonstrate that COOL has superior performance compared to the existing algorithms.

Sparse Mobile CrowdSensing (SMCS) is an emerging distributed data collection framework. As one of the core methods,data completion uses the collected data to fill in the missing data. However, this approach inevitably poses significant privacy risks, since the traditional data completion methods require users' time and location information. In this paper, we propose a federated learning-based distributed data completion framework, which employs matrix factorization (MF) for local completion model training and federated learning to aggregate parameters of the MF model. This enables the construction of a global completion model without requiring private data, thereby mitigating privacy concerns. To address the challenges posed by the sparsity and asynchrony of distributed data, we incorporate time-aware deep neural network architectures with an asynchronous mechanism to enable federated training under irregularly gathered local data. Experimental results demonstrate that the proposed model achieves high data completion accuracy while ensuring robust privacy protection.

Decentralized machine learning (DML) supports collaborative training in large-scale networks with no central server. It is sensitive to the quality and reliability of inter-device communications that result in time-varying and stochastic topologies. This paper studies the impact of unreliable communication on the convergence of DML and establishes a direct connection between the spectral properties of the mixing process and the global performance. We provide rigorous convergence guarantees under random topologies and derive bounds that characterize the impact of the expected mixing matrix's spectral properties on learning. We formulate a spectral optimization problem that minimizes the nontrivial spectral radius of the expected second-order mixing matrix to enhance the convergence rate under probabilistic link failures. To solve this non-smooth spectral problem in a fully decentralized manner, we design an efficient subgradient-based algorithm that integrates Chebyshev-accelerated eigenvector estimation with local update and aggregation weight adjustment, while ensuring symmetry and stochasticity constraints without central coordination. Experiments on a realistic low Earth orbit satellite constellation with time-varying inter-satellite link models and real-world remote sensing data demonstrate the feasibility and effectiveness of our method. The method significantly improves classification accuracy and convergence efficiency compared to existing baselines, validating its applicability in satellite and other decentralized systems.

Sparse crowdsensing tackles data incompleteness through sparse sensing coupled with inference-based completion. Yet prevailing solutions face three inherent constraints: (i) Spatiotemporal completion methods demand extensive historical/cross-domain data for source-domain pretraining and target-domain fine-tuning; (ii) Such dependency causes catastrophic performance degradation in cold-start scenarios lacking target-domain samples, a setting that remains largely unaddressed by existing data-dependent frameworks; (iii) Resource-intensive pretraining-fine-tuning pipelines hinder lightweight deployment. To overcome these challenges, this paper proposesMappingTransfer Learning-basedSpatiotemporalCompletion method (MapT-STC), a one-shot spatiotemporal completion framework that advances beyond few-shot approachesCorrelatedDataFusion forMatrix Completion (CDFMC) by featuring a remodeled pretraining-finetuning paradigm. Departing from conventional approaches, MapT-STC implements a dual-path learning system: it simultaneously trains a target-specialized model for domain-specific features and a cross-domain generalizable model viaCorrelationAlignment (CORAL). The framework explicitly aligns feature distributions by decomposing spatiotemporal data into separable temporal patterns and spatial structures, thereby enhancing the quality of domain adaptation. Ultimately, Kalman-filter-driven fusion integrates predictions from both specialized and generalized models for optimal completion. Experiments demonstrate MapT-STC's superiority under cold-start conditions: reducing initial error by over 30% and accelerating convergence time by at least 7.4% compared to state-of-the-art baselines, using only one target-domain training sample.

Cooperative perception is a crucial technology to improve the safety and robustness of autonomous driving systems. Existing works mainly utilize a pre-constructed 3D map or roadside infrastructures to estimate poses for cooperative perception. However, the pre-constructed map should be updated frequently, and roadside infrastructures can only be employed on specific roads. To overcome these limitations, we exploit the inherent features of external shapes of vehicles for relative pose estimation and design an Edge-assisted multi-Vehicle cOoperative perception system, called EVO. EVO adopts a suit of approaches to ensure the accuracy and real-time performance of cooperative perception: (1) a novel approach is proposed to estimate relative poses without requiring any pre-constructed 3D map or roadside infrastructures; (2) a hierarchical registration module is adopted to speed up pose estimation; (3) a Regions of Interest (RoIs) prediction module based on Dlinear and a registrations reduction module are designed to reduce the computation latency. We evaluate EVO on a CARLA-based dataset and demonstrate that EVO can perform multi-vehicle cooperative perception in real-time and maintain centimeter-level accuracy in pose estimation. Further, EVO extends the maximum perception distance from 32-39m to 38-67m and achieves about 2.43× improvement in perception area.

Ensuring stable and high-quality real-time video analytics for computationally constrained mobile agents is essential. However, limited computing resources and network bandwidth present significant challenges in meeting the objective of low response time and high inference accuracy. In this paper, we present DiVE, an edge-assisted video analytics system that utilizes motion vectors calculated by video codec to extract foregrounds and differentially encode frames. DiVE removes rotational components from motion vectors by solving over-determined linear equations and filters noisy motion vectors based on the observation that motion vectors of static objects point to the same point when the ego agent purely translates. To distinguish foregrounds from backgrounds, DiVE estimates the ground based on observations that all foregrounds stand on the ground and motion vectors on static objects at the same height have the same normalized magnitude. DiVE then uses region-growing-based clustering to identify foreground objects. An adaptive bitrate allocation method is applied to optimize accuracy under estimated bandwidth. We implement a prototype and conduct extensive experiments to evaluate the performance of DiVE. The results demonstrate that DiVE can improve detection accuracy by up to 19.0% and reduce response time by up to 56.0% compared with other video analytics schemes in real-world traces.

Cyclic fault diagnosis is crucial for ensuring system reliability, as it facilitates the early detection and resolution of recurring failures, minimizing downtime and maintenance costs. Existing methods often face challenges with high computational complexity and scalability, limiting their real-time applicability in complex networks like Bijective Connection (BC) networks, where rapid fault isolation is essential. In this paper, we explore the cyclic diagnosability ofg-BC networks under two classical system-level fault diagnosis models: the P/M/C model and the MM* model. Cyclic diagnosability, which focuses on maintaining connectivity with cycles in the residual subgraph after failures, is a more robust measure than traditional diagnosability. We prove that for anyg-BC networkg-Xnthat satisfies Definition 3, the cyclic diagnosability is given byct(g-Xn) = 5n− 8 −gforg= 1,n≥ 12 and 2 ≤g≤ 5,n≥ 11 under both models, thereby determining the maximum number of faults that can be accurately identified while preserving at least two cyclic components. To support practical diagnosis under these models, we propose two fast and scalable algorithms with low time complexity: TDPMC (Threshold-Based Fault Diagnosis under the P/M/C Model forg-BC Networks) and FBDMM (Suspicion-Score-Based MM* Fault Diagnosis with Dynamic Threshold forg-BC Networks). Both methods exploit the topological features of BC networks to efficiently identify cyclic faults. The time complexities of TDPMC and FBDMM areO(nN) andO(n2N), respectively. Experimental results on an 11-dimensional and 12-dimensionalg-BC networks validate the theoretical findings, demonstrating significant improvements in accuracy, recall, and fault detection reliability.

The Internet of Things (IoT) is gathering paces in the new era of Industry 4.0, and the Digital Twin (DT) technology bridges the gap between the bursting amounts of data generated by IoT devices and the user requirements for real-time data processing. DT services maintain living digital models of physical objects, and a DT network enables comprehensive service provisioning with the global knowledge of a group of DTs. On the other hand, exposing serverless computing at network edges, the recent advances in Mobile Edge Computing (MEC) introduce new inspirations to the DT landscape that ensure fine-grained resource management and low network-wide delay of DT services. However, social relationships among IoT devices and DT data privacy impact DT orchestrations. In this paper, we first design a differential privacy-based federated learning framework to build a DT network for DT services in response to user requests in an MEC, thereby enhancing the Quality of Services (QoS). Built upon the proposed framework, we then formulate two novel social-aware DT placement problems: the static social-aware S_DT placement problem, and the dynamic social-aware S_DT placement problem, respectively. We also show the NP-hardness of the defined problems. Then, we formulate an Integer Linear Program (ILP) solution to the static social-aware S_DT placement problem when the problem size is small; otherwise we develop an approximation algorithm with a provable approximation ratio for it. Third, we study the dynamic social-aware S_DT placement problem when requests arrive one by one without the knowledge of future request arrivals over the time horizon, for which we devise an online algorithm with a provable competitive ratio. Finally, we conduct simulations to evaluate the performance of the proposed algorithms. Simulation results show that the proposed algorithms outperform their counterparts, improving the performance compared with their baselines by no less than 14.9%.

As the core infrastructures in the information systems, the data center networks carry a large number of tasks of data processing and storage. In a data center network, intermittent faults are often difficult to be found and dealt with in time because of their hiddenness and uncertainty. Once these faults accumulate to a certain extent, they can cause serious network outages and even lead to the collapse of the entire data center. In order to discover and resolve these potential problems in a timely manner, it is crucial to apply intermittent fault diagnosis, thus ensures the continuous and stable operation of the data center. In this paper, we propose the intermittent fault diagnosabilitytPMCI(Cn) for ann-dimensional data center network CSDC under the Preparata/Metze/Chien model (PMC model) by establishing the fault tolerance of the network. Additionally, under the PMC model, we propose a probabilistic multiple intermittent fault diagnosis algorithm (PMIFDPMC) with time complexityO(nN) by preferentially generating weighted multiple test networks (GWMTN) whereNis the scale of CSDC. Moreover, we apply the algorithm PMIFDPMC to a 7-dimensional CSDC and a real-world dataset of the Internet of Things. Across different scenarios of intermittent fault nodes, we calculate the Accuracy, Recall, FNR, G-mean, and F1-score using various testing iterations. The experimental results demonstrate that, as the number of testing iterations of algorithm PMIFDPMC increases, the quantity of intermittent fault nodes that are correctly diagnosed also increases. This highlights the favorable performance and effectiveness of algorithm PMIFDPMC on the real-world dataset of the Internet of Things.

Federated Learning (FL) enhances data privacy for End Equipment Workers (EWs) by enabling the sharing of model parameters instead of raw data. However, energy constraints and individual self-interest may discourage EWs from participating or slow down training, ultimately affecting the performance of the global FL model. To address these challenges, we propose a three-stage Stackelberg game-based framework that leverages wireless power to incentivize participation while ensuring the successful completion of FL tasks. In this framework, the Base Station (BS) publishes FL task and seeks to obtain an improved global model at a reduced cost. EWs train local models, aiming to maximize their payments while minimizing energy consumption. Meanwhile, the Charging Service Provider (CSP) supplies energy to EWs via Wireless Power Transfer (WPT) during model training and uploading, charging appropriate fees for the service. We employ the backward induction method to analyze the proposed game problem, proving the existence of a unique Stackelberg equilibrium and Nash equilibrium. Furthermore, we propose the Trust Region Method (TRM) to solve the unit payment strategy problem of BS. Extensive simulations validate that our method consistently outperforms benchmark schemes, achieving higher average utility across a wide range of scenarios.

Distributed training of large language models demands efficient gradient synchronization strategies. Existing All Reduce algorithms often fail to fully leverage underlying topological properties, resulting in low edge utilization and unbalanced load in hypercube networks. This paper proposes a Redundant Hierarchical Ring All-Reduce (RHRA) algorithm tailored for hypercube topologies. The algorithm first designs a hierarchical Ring All-Reduce structure with high edge utilization by exploiting the hypercube's inherent high parallelism and symmetry, proving that the optimal number of hierarchical layers is n 2. It then introduces controlled redundancy, dynamically adjusting transmission strategies based on the relationship between the value of bandwidth B and the value of data volume D to reduce communication hops and transmission latency while ensuring reliability. Experiments demonstrate that our algorithm achieves the minimum number of gradient synchronization hops under varying bandwidth constraints, with end-to-end transmission time significantly outperforming existing schemes. Further reliability simulations verify its effectiveness in balancing redundancy to prevent resource waste while maintaining high reliability.

Sparse Mobile CrowdSensing (MCS) is a cost-effective paradigm which recruits users to sense only a small amount of data and infer the rest ones. The existing works usually provide rewards for users to cover their sensing costs. Compared with the costs, we actually need to pay for the value of data, which is a new and valuable perspective of data pricing. However, in Sparse MCS, we can hardly estimate the value of data accurately, as it depends not only on the data itself but also on its contribution to inference. Moreover, the sensing data is naturally noisy and its spatiotemporal information may be invisible until we pay for it, which makes the data pricing more challenging. To this end, we investigate the value of sensing data, and propose a data pricing framework for Sparse MCS. We first present the sampling-based uncertainty quantification to evaluate the value of data, with the aim to pay for the intrinsic data values but not the indirect costs. As a byproduct, the noise can be dealt with naturally by such sampling-based methods. Further considering the availability of spatiotemporal information, we also study three cases of data pricing, i.e, paying for sensed data, spatiotemporal information, and next data. Extensive experiments on five real-world data sets verify the effectiveness of our proposals.

As mobile edge computing technology advances rapidly, latency-sensitive and resource-intensive applications are being offloaded to edge servers to enhance Quality of Service (QoS) for users. Traditional monolithic architectures, however, struggle to meet the escalating service and traffic requirements of distributed users due to their inherent inflexibility. In response to these challenges, microservices architecture, characterized by scalability and flexibility, has been adopted for dynamic deployment at the network edge. However, the deployment of these lightweight, dependency-rich components in a way that minimally impacts the makespan and maximizes quality of service is complex. Current studies often overlook the deployment of microservices with specific dependencies within constrained environments of edge server clusters and communication links. This paper introduces practical and effective strategies for the deployment and updating of microservices, tailored to various application contexts. Initially, two scenarios are analyzed: one constrained by bandwidth with unlimited storage, and the other by storage with unlimited bandwidth. For each scenario, optimal solutions are developed using a novel enhanced graph construction method. The study progresses to a more intricate scenario involving comprehensive constraints on storage, computation, and communication resources. An optimized deployment method is proposed, utilizing main path embedding followed by an innovative simulated annealing algorithm for iterative refinement. This method is validated by demonstrating that the main path coincides with the critical path. Furthermore, the dynamic reallocation of edge resources is explored through a critical path-based updating algorithm that optimizes microservice locations to reduce overall makespan. Extensive experiments demonstrate that our strategies outperform existing representative benchmark approaches in terms of overall performance and microservice deployment efficiency.

Serverless workflows introduce unique challenges for modern cluster schedulers, as they consist of highly concurrent and ephemeral functions with unpredictable execution patterns. Through analysis of workloads derived from production serverless trace characteristics, we observe that existing OS-level schedulers, such as Linux CFS, lack workflow-level awareness and make scheduling decisions solely at the function level, which can result in bottlenecks within workflows and prolonged Workflow Completion Times (WCTs). We present Metis, anon-clairvoyant, workflow-aware OS scheduler designed specifically for serverless workflow workloads, which aims to reduce average WCTs by treating workflows as first-class scheduling entities. Metis implements Workflow-Aware Least-Attained Service (WLAS), a non-clairvoyant scheduling algorithm that leverages workflow-level virtual clocks and critical path estimation to reduce WCTs and ensure fairness. By utilizing eBPF to hook into existing OS primitives, Metis achieves practical deployment with minimal kernel modifications. Extensive synthetic trace-driven simulations demonstrate that Metis reduces average WCTs by 31.3% and the 95th percentile by 14.3% compared to state-of-the-art function-centric scheduling approaches. Real-system experiments across diverse workflow patterns show improvements ranging from 47.2% to 58.2% in average end-to-end latency and the 99th percentile by up to 73.9% over baseline schedulers.

While GNN-based services often experience load fluctuation, applying serverless computing to serve GNN inference reduces the cost and allows elastic resource scaling. However, GNN serverless shows poor performance due to heavy data fetching latency and long cold startup overhead, and our observation indicates opportunities for reducing data redundancy and mitigating cold startup latency. In this paper, we present FaaSGNN, a serverless GNN inference framework that enables low latency and memory efficient GNN serving through three key designs: (i) serverless-native on-demand graph fetching strategy that enables lightweight in-container graph sampling with full dataset resides in remote; (ii) memory-aware adaptive feature caching policy, which facilitates data reuse between requests to reduce redundant fetching; and (iii) load-aware request scheduler, which reschedules requests to bypass cold start and achieve load balance between containers. Experimental results show that FaaSGNN achieves a 5.6x lower end-to-end latency and 57.1% less memory usage on average compared to state-of-the-art works.

Microservices have become an important design paradigm for large-scale distributed systems, offering flexible provisioning options. A fundamental challenge is the exponential growth of the solution space with the number of user requests, posing challenges to efficient provisioning and scheduling when aiming to balance cost and latency under resource constraints in large-scale dynamic edge environments. To tackle this problem, we formulate a joint optimization model for microservice provisioning and routing that integrates cost efficiency and latency reduction while accounting for uncertainties in the origin location of requests. To establish a unified framework that facilitates decision-making, we propose an integer linear programming (ILP) model that captures the dependencies between microservices in the service chain. Our Scalable optimization framework with Cost-efficiency and Latency reduction (SoCL) comprises three stages: an initial partitioning guarantees latency bounds, a pre-provisioning stage considers provisioning cost, and a multi-scale combination stage balances cost and latency through parallel and serial local search. Extensive experiments conducted across diverse scenarios based on a commonly used dataset demonstrate that the proposed SoCL framework significantly increases cost efficiency and decreases latency compared to established baselines, while reducing execution time up to one order of magnitude compared to obtaining the optimal solution by optimizer.

Large Language Models (LLMs), constrained by limited context windows, often face significant performance degradation when reasoning over long contexts.To address this, Retrieval-Augmented Generation (RAG) retrieves and reasons over chunks but frequently sacrifices logical coherence due to its reliance on similarity-based rankings.Similarly, divideand-conquer frameworks (DCF) split documents into small chunks for independent reasoning and aggregation.While effective for local reasoning, DCF struggles to capture longrange dependencies and risks inducing conflicts by processing chunks in isolation.To overcome these limitations, we propose ToM, a novel Tree-oriented MapReduce framework for long-context reasoning.ToM leverages the inherent hierarchical structure of long documents (e.g., main headings and subheadings) by constructing a DocTree through hierarchical semantic parsing and performing bottom-up aggregation.Using a Tree MapReduce approach, ToM enables recursive reasoning: in the Map step, rationales are generated at child nodes; in the Reduce step, these rationales are aggregated across sibling nodes to resolve conflicts or reach consensus at parent nodes.Experimental results on 70B+ LLMs show that ToM significantly outperforms existing divide-andconquer frameworks and retrieval-augmented generation methods, achieving better logical coherence and long-context reasoning.Our code is available at https://github.com/gjn12- 31/ToM.

Jie Wu, Haoling Li, Xin Zhang, Xiao Liu, Yangyu Huang, Jianwen Luo, Yizhen Zhang, Zuchao Li, Ruihang Chu, Yujiu Yang, Scarlett Li. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

The growing demands of compute-intensive applications have led to the emergence of Computing Power Networks (CPNs), which integrate diverse computing resources across cloud, edge, and devices for efficient task scheduling and resource allocation in a network environment. This paper addresses the critical challenge posed by the dual spatiotemporal dynamics of tasks and resources, which complicates the optimization process of task scheduling in CPNs. Our work is the first to systematically investigate the joint optimization of task scheduling and traffic routing in green CPNs involving the dual dynamics. We formulate the profit maximization problem as an integer nonlinear programming problem that is NP-hard. To tackle this issue, we propose a novel deep reinforcement learning based algorithm capable of SpatioTemporal Dynamic Perception (STPer), which employs Graph Neural Networks(GNN) and Long Short-Term Memory(LSTM) networks. The STPer algorithm effectively captures the spatiotemporal dynamics of both resources and tasks, maximizing platform profits. Extensive simulations demonstrate that STPer significantly outperforms existing benchmark algorithms, highlighting its superior performance in optimizing resource utilization and enhancing profitability in the green CPN.

Real-time video analytics (VA) presents challenges due to its computational intensity and latency sensitivity, especially when processed on mobile devices with limited local resources. We propose to offload VA tasks to edge servers with diverse computational capabilities. We present a "detect + track" approach with on-device object tracking and edge-assisted object detection. We formulate a long-term nonlinear integer programming to maximize the overall accuracy within detection frequency and latency constraints. We then design a queue-based online optimization algorithm to solve it: relax the original problem from the integer domain to the real domain, then employ a queue-based adaptation and randomized rounding strategy. Via rigorous proof, both dynamic regret regarding detection accuracy and the real-time requirement are ensured. Evaluation results also demonstrate the effectiveness of our approach.

The demand for video analytics is increasing rapidly. Due to the limited computational and network resources on edge servers, adjusting video configurations such as resolution and frame rate has become an effective strategy to reduce computational and transmission costs. However, this can also compromise detection accuracy, necessitating a balance between resource consumption and analytics accuracy. Also, the dynamic availability of edge servers and variability in their energy consumption further complicates making offloading decisions and configuration selection. In this paper, we first model the problem as a mixed planning program. Then we propose a volatile MAB-based configuration selection algorithm, VACS, which aims to maximize video analytics accuracy while reducing the overall energy consumption. Rigorous proof measures the gap between online decisions and the optimum. Extensive experiments validate the effectiveness of VACS.

Graph computing in cloud-edge collaborative environments faces critical challenges in distributed task processing, particularly in fundamental operations such as subgraph isomorphism that underpins triangle counting applications. In typical architectures where data streams are transmitted from edge collectors to cloud masters, conventional approaches employ reservoir sampling to distribute edge streams among workers for triangle estimation. However, the computational accuracy degradation is caused by cross-domain edge distribution strategies. In this paper, we propose a cloud-edge collaborative framework for distributed triangle counting. We employ spectral clustering analysis to reveal latent domain relationships that guide edges distribution. Our experimental evaluation uses streaming data with global relative error measurement across multiple datasets, demonstrating superior performance over existing algorithms.

Content delivery in vehicular edge networks faces critical challenges due to dynamic user mobility, unpredictable content request patterns, and limited storage at edge nodes. To tackle these problems, we propose a distributed online framework that jointly performs proactive caching at roadside units (RSUs) and hierarchical prefetching from the cloud to macro base stations (MBSs), enabling real-time adaptation to spatiotemporal variations in content demand across different time scales. Our goal is to minimize content transmission latency while satisfying system-wide resource and cost constraints. The proposed Vehicular-based Online Proactive caching and Prefetching (VOPP), integrates trajectory-based user mobility prediction with future content demand estimation to guide online distributed caching. At the RSU level, we formulate a distributed online convex optimization model with fine-grained gradient updates and inter-agent coordination based on real-time mobility patterns. At the MBS level, we construct a multi-step predicted content set using user mobility and request forecasts, and define a value density metric that combines popularity and delay reduction. On both levels, additional subsequent refinement steps ensure high-quality caching decisions. Extensive simulations based on a real-world GPS dataset of 10,357 taxi trajectories in Beijing demonstrate that VOPP significantly reduces transmission delay and achieves robust performance across diverse mobility patterns and user densities, outperforming baseline methods.

Radio frequency (RF) fingerprinting is a technique used to identify a wireless device based on its specific and unique hardware characteristics. In recent years, deep learning has been utilized for RF fingerprinting due to its superiority in feature extraction and higher classification accuracy. However, one major challenge of deep learning-based RF fingerprinting is that wireless signals are highly sensitive to environmental conditions, causing the device fingerprints captured in one environment to not transfer well to another. Hence, deep learning models are found to perform well in the same condition but lose their ability to classify devices in the new condition. In this paper, we examine three transfer learning techniques to mitigate the domain shift problem in RF fingerprinting and compare them with two well-defined baselines. The three RF fingerprinting datasets under various scenarios are examined to explore how environmental factors impact RF fingerprinting, such as transmitter locations, transmitter distance, and device configurations. We identify the most challenging scenarios and study how environmental factors lead to model deterioration through t-SNE visualization.

Text-to-video models have made remarkable advancements through optimization on high-quality text-video pairs, where the textual prompts play a pivotal role in determining quality of output videos. However, achieving the desired output often entails multiple revisions and iterative inference to refine user-provided prompts. Current automatic methods for refining prompts encounter challenges such as Modality-Inconsistency, Cost-Discrepancy, and Model-Unaware when applied to text-to-video diffusion models. To address these problem, we introduce an LLM-based prompt adaptation framework, termed as Prompt-A-Video, which excels in crafting Video-Centric, Labor-Free and Preference-Aligned prompts tailored to specific video diffusion model. Our approach involves a meticulously crafted two-stage optimization and alignment system. Initially, we conduct a reward-guided prompt evolution pipeline to automatically create optimal prompts pool and leverage them for supervised fine-tuning (SFT) of the LLM. Then multi-dimensional rewards are employed to generate pairwise data for the SFT model, followed by the direct preference optimization (DPO) algorithm to further facilitate preference alignment. Through extensive experimentation and comparative analyses, we validate the effectiveness of Prompt-A-Video across diverse generation models, highlighting its potential to push the boundaries of video generation.

Along with the advance of low-altitude economy, a variety of applications based on Unmanned Aerial Vehicles (UAVs) have been developed to accomplish diverse tasks. In this paper, we focus on the scenario of multiple UAVs performing Federated Learning (FL) tasks. Specifically, a group of UAVs is scheduled to repeatedly visit some Points of Interest (PoIs), collect the data produced by these PoIs, and jointly train a machine learning model based on the collected data. The most challenging issue is how to schedule UAVs to collect data so as to optimize the generalization and convergence of model training under the case that the distributions of the data produced by PoIs have not been known in advance. To address this issue, we propose a novel framework for online FL on distributed unknown data, named OFL-UD2, which is dedicated to online decision-making for UAVs to optimize model training performance. Concretely, we formulate the optimization problem while considering the convergence and quality of trained models as well as energy constraints. Then, we define a utility metric for the data quality of different PoIs and conduct a rigorous convergence analysis for OFL-UD2. Based on the analysis results, we design a two-stage algorithm to determine the scheduling of UAVs. Extensive simulations demonstrate that OFL-UD2can improve model accuracy and speed up running time compared to existing benchmarks significantly.

The rapid advancement of Large Vision-Language models (LVLMs) has demonstrated a spectrum of emergent capabilities. Nevertheless, current models only focus on the visual content of a single scenario, while their ability to associate instances across different scenes has not yet been explored, which is essential for understanding complex visual content, such as movies with multiple characters and intricate plots. Towards movie understanding, a critical initial step for LVLMs is to unleash the potential of character identities memory and recognition across multiple visual scenarios. To achieve the goal, we propose visual instruction tuning with ID reference and develop an ID-Aware Large Vision-Language Model, IDA-VLM. Furthermore, our research introduces a novel benchmark MM-ID, to examine LVLMs on instance IDs memory and recognition across four dimensions: matching, location, question-answering, and captioning. Our findings highlight the limitations of existing LVLMs in recognizing and associating instance identities with ID reference. This paper paves the way for future artificial intelligence systems to possess multi-identity visual inputs, thereby facilitating the comprehension of complex visual narratives like movies.

Existing methods for code generation use code snippets as seed data, restricting the complexity and diversity of the synthesized data. In this paper, we introduce a novel feature tree-based synthesis framework, which revolves around hierarchical code features derived from high-level abstractions of code. The feature tree is constructed from raw data and refined iteratively to increase the quantity and diversity of the extracted features, which captures and recognizes more complex patterns and relationships within the code. By adjusting the depth and breadth of the sampled subtrees, our framework provides precise control over the complexity of the generated code, enabling functionalities that range from function-level operations to multi-file scenarios. We fine-tuned widely-used base models to obtain EpiCoder series, achieving state-of-the-art performance on multiple benchmarks at both the function and file levels. In particular, empirical evidence indicates that our approach shows significant potential in the synthesizing of repository-level code data. Our code and data are publicly available.

LoRa enables long-range Internet of Things (IoT) connectivity but suffers from collision issues in dense deployments, where concurrent transmissions overlap at gateways, leading to decoding errors. Existing solutions rely on time/frequency separation or protocol modifications, requiring either hardware changes or dedicated codings, and failing to resolve collisions that are completely aligned in time or frequency. We present CD-LoRa, a Channel-Division based LoRa parallel transmission scheme for LoRa collisions. CD-LoRa exploits distinct channel signatures as inherent orthogonal fingerprints, enabling parallel decoding even when collisions are completely aligned. We present a phase calibration model that decouples genuine channel features from hardware imperfections and payload modulation distortions. We enhance low-SNR LoRa signals through energy-concentration processing. We address channel variations in mobile scenarios with a dynamic temporal sequence based clustering design. We implement CD-LoRa on commodity LoRa devices and evaluate its performance in real-world deployments. Experimental results show that CD-LoRa effectively decodes up to 8 time-frequency-aligned packets, and improves network throughput by 1.72× compared to state-of-the-art methods.

Edge Video Analytics (EVA) significantly reduces response time by executing analytical tasks at the edge. However, it inevitably faces accuracy loss when dealing with highly complex analytical scenarios. To overcome this, we propose offloading the most complex video frames to the cloud while processing other frames at the edge. Nevertheless, determining both the quantity and the specific selection of frames for offloading poses challenges due to edge-cloud bandwidth constraints and the dynamic nature of video content. To tackle this problem, we present a Discerning and Elastic Offloading Framework (DEOF), which consists of an Accuracy Predictor and an Offloading Scheduler. The former identifies the detection complexity of each frame by predicting its F1-score gain based on multidimensional information, enabling it to discern and select the most complex frames for offloading. The latter determines the optimal proportion of frames to process in the cloud and at the edge by designing a Lyapunov-optimization-based algorithm, which elastically adjusts this proportion in response to time-varying video content and resource conditions, thus ensuring both adaptability and efficiency. We have implemented DEOF fully based on COTS hardware, and the experimental results demonstrate the effectiveness of DEOF, showing that our system can reduce offloaded data volume by$7.1 \%-36.3 \%$, decrease latency by$\mathbf{2. 6 \% - 1 9. 5 \%}$, and improve accuracy by$\mathbf{2. 6 \%}$3.2 % compared to alternative methods.

With the proliferation of edge computing and the Internet of Things (IoT), inference-driven intelligent cameras are increasingly deployed for resource-efficient and privacypreserving processing. In real-world scenarios, such as traffic surveillance, cameras deployed at different locations (e.g., intersections or corners) experience imbalanced inference workloads, leading to latency bottlenecks and resource underutilization. To address this, we propose ReMO, an adaptive framework for collaborative video analytics. Unlike full-frame offloading, ReMO divides video frames into regions to reduce data transmission and enable fine-grained load balancing. It consists of two key components: a Region Generator that analyzes scene features to identify regions with varying detection needs, and a Region Scheduler that formulates scheduling as an integer nonlinear problem, solved via an adaptive online algorithm based on Lyapunov optimization and Markov approximation. Experimental results demonstrate that ReMO effectively reduces latency by$9.3-26.3 \%$while maintaining high accuracy, outperforming baseline strategies.

Modern distributed file systems (DFSs) rely on metadata server clusters to manage large-scale files and achieve scalability. However, the hierarchical namespace structure and dynamic user workloads pose severe challenges for efficient metadata partitioning and load balancing. Existing approaches primarily focus on identifying and redistributing hot metadata to address imbalances. While these load-balancing strategies offer potential benefits, they often reduce metadata locality, ultimately failing to improve the end-to-end job completion time—a key metric prioritized by users. Although recent research reveals that learning-based approaches are effective in predicting hotspots, they have been shown to be less effective in improving metadata performance. We revisit metadata load balancing strategies and propose a learning-based metadata load balance framework Origami, which focuses on minimizing end-to-end job completion time rather than equalizing loads. Origami first decomposes the overhead of metadata operations and assesses the impact of migration decisions on user requests, allowing us to compute the benefits of migration decisions for job completion time when future requests are known. Subsequently, Origami propose the Meta-OPT algorithm to determine near-optimal migration decisions. Finally, we implemented OrigamiFS, on which we collected statistical data to train and validate ML-models capable of predicting migration benefits. By predicting the benefits of migration decisions and employing Meta-OPT to quickly explore nearly optimal migration decisions, Origami makes a better trade-off between load balancing and namespace locality. Our evaluation shows that compared to state-of-the-art methods, Origami increases aggregated metadata throughput by 1.12-2.51 × across three real-world workloads, and enhances end-to-end throughput by 1.11-2.02 ×.

The rapid growth of deep learning has created unprecedented demand for computing resources, while many small and enterprise-level clusters remain underutilized. Computing resource exchange platforms offer a solution by aggregating these idle resources. However, effective cluster-task matching depends on accurate performance prediction. Existing approaches, which decouple prediction from matching, often lead to suboptimal decisions due to misaligned objectives. We propose a Matching-Focused Cluster Performance Predictor (MFCP), an end-to-end framework that integrates performance prediction with task matching to improve decision accuracy and resource utilization. Unlike existing methods that prioritize prediction accuracy, MFCP minimizes decision regret by aligning the predictor’s loss with optimal matching objectives. To handle non-differentiable matching optimization, we use continuous relaxation and incorporate constraints via an interior-point method, ensuring meaningful gradients for training. For non-convex optimization, we approximate optimal decisions with gradient descent and estimate gradients using zeroth-order perturbation. Experiments show that MFCP consistently outperforms existing methods across different cluster environments and scales, achieving lower matching regret and higher resource utilization.

The emergence of byte-addressable memory technologies, such as CXL-attached memory, has catalyzed extensive research into tiered memory management. Existing tiering solutions optimize system-wide performance by migrating frequently accessed (“hot”) data to fast-tier memory via page migration, which serves as the de facto mechanism in modern OS. However, these strategies often fail in the multi-tenant environment, where diverse workloads interfere with each other. For example, latency-critical workloads co-located with throughput-oriented ones may face the “cold page dilemma,” where critical pages are misclassified as “cold” and migrated to slower tiers, leading to significant performance degradation. Moreover, current methods often assume negligible migration overhead, which becomes problematic in multi-core systems handling write-intensive workloads that incur substantial costs. This paper proposes Vulcan, a workload-aware tiered memory management framework that targets fair and efficient tiering in multi-tenant environments. Vulcan introduces four key innovations: (1) workload-dependent migration mechanism, which decouples page migration from the OS kernel to enhance operational flexibility for multi-workloads; (2) QoS-aware fair resource partitioning, which dynamically optimizes fast memory distribution through per-workload fast tier hit ratios and fairness-oriented allocation policies; (3) per-thread page table replication, which minimizes TLB coherence overhead during migration; and (4) biased page migration policy, which optimizes efficiency by considering both access characteristics (read-intensive vs. write-intensive) and thread-level page ownership (private vs. shared). We evaluated Vulcan using multiple representative cloud applications with realistic working sets in co-location scenarios. Vulcan improves performance by 12.4% on average and achieves a 75.3% improvement in fairness compared to existing state-of-the-art memory tiering solutions.

Federated learning is vulnerable to model poisoning attacks in which malicious participants compromise the global model by altering the model updates. Current defense strategies are divided into three types: aggregation-based methods, validation dataset-based methods, and update distance-based methods. However, these techniques often neglect the challenges posed by device heterogeneity and asynchronous communication. Even upon identifying malicious clients, the global model may already be significantly damaged, requiring effective recovery strategies to reduce the attacker's impact. Current recovery methods, which are based on historical update records, are limited in environments with device heterogeneity and asynchronous communication. To address these problems, we introduce FedHAN, a reliable federated learning algorithm designed for asynchronous communication and device heterogeneity. FedHAN customizes sparse models, uses historical client updates to impute missing parameters in sparse updates, dynamically assigns adaptive weights, and combines update deviation detection with update prediction-based model recovery. Theoretical analysis indicates that FedHAN achieves favorable convergence despite unbounded staleness and effectively discriminates between benign and malicious clients. Experiments reveal that FedHAN, compared to leading methods, increases the accuracy of the model by 7.86%, improves the detection accuracy of poisoning attacks by 12%, and enhances the recovery accuracy by 7.26%. As evidenced by these results, FedHAN exhibits enhanced reliability and robustness in intricate and dynamic federated learning scenarios.

Federated learning, as a distributed learning framework, aims to develop a global model while preserving client privacy. However, heterogeneity of client data leads to fairness issues and reduced performance. Techniques like parameter decoupling and prototype learning appear promising, yet challenges such as forgetting historical data and limited generalization persist. These methods also lack local insights, with locally trained features prone to overfitting, which affects generalization in global parameter aggregation. To address these challenges, we propose FedCPD, a personalized federated learning framework. FedCPD maintains historical information, reduces information loss, and increases personalization through hierarchical feature distillation and cross-layer feature fusion. Moreover, we utilize representation techniques like prototype contrastive learning and prototype alignment to capture diverse client data features, thus improving model generalization and fairness. Experiments show FedCPD outperforms state-of-the-art models, enhancing generalization by up to 10.40% and personalization by up to 4.90%, highlighting its effectiveness and superiority.

Model inversion and adversarial attacks in semantic communication pose risks, such as content leaks, alterations, and prediction inaccuracies, which threaten security and reliability. This paper introduces, from an attacker's viewpoint, a novel framework called RepObE (Representation Learning-Enhanced Obfuscation Encryption Modular Semantic Task Framework) to secure semantic communication. This framework employs dynamic encryption during semantic extraction and feature transmission to hinder attackers from reconstructing data through eavesdropping, thus strengthening system privacy. To combat image communication task challenges, we propose a prototype adversarial collaborative alignment training approach enhanced by representation learning. This method extracts and encodes semantic features while using dynamic perturbation and robust optimization to improve system resilience against adversarial threats. The approach ensures reliable semantic communication in complex environments, maintaining performance while countering attacks using feature obfuscation, adversarial training, and representation learning. Experimental results demonstrate that our method surpasses existing techniques by more than 2% in resisting model inversion attacks on classification tasks. Visually, our method excels with minimal decipherable images for attackers. It also shows a 3% to 5% improvement in countering adversarial attacks on classification tasks.

Current Post-Training Quantization (PTQ) schemes have been extensively studied for traditional convolutional neural networks and language models; however, PTQ application in diffusion models has shown significant performance degradation due to static settings of PTQ. Existing methods only uniformly and statically sample during each denoising step to construct calibration sets, neglecting the different importance of different steps in diffusion models. Furthermore, diffusion models exhibit a large number of activations with skewed distributions, and maintaining a static zero-point during the reconstruction process causes the model to converge only to local optima. To solve these limitations, it is necessary to dynamically design calibration dataset construction methods for different quantization scenarios and develop specialized optimization strategies tailored to specific activation distributions. Thus we proposed a unified framework, termed Dynamic PTQ, to achieve the aforementioned purposes. The framework first applies an evolutionary search algorithm to dynamically construct calibration sets for different quantization scenarios. Then, we design a dynamic zero-point update strategy for the quantizer, significantly reducing the loss during the reconstruction process. Extensive experiments demonstrate that our method outperforms current PTQ methods for diffusion models in generating high-quality samples. In particular, for the LSUN-bedrooms 256×256 task, our method quantizes the corresponding full-precision LDM-4 to W4A6 with only a 0.84 increase in FID.

Sparse Mobile CrowdSensing has emerged as a practical method for data collection, recruiting mobile users to collect partial data and leveraging spatiotemporal correlations to infer the missing data. To improved the QoS of crowdsourced data, existing methods typically assume that the scales of collected data are similar. However, in real-world scenarios, the diversity of user devices results in data collections that vary in scale. More importantly, the collected coarser-scale data points often do not perfectly correspond to multiple finer-scale data points, resulting in highly complex compositional relationships and posing significant challenges for multi-scale data completion. To address these challenges, this paper proposes a multi-scale data completion framework designed to process and integrate multi-scale data with non-aligned compositional relationships. We first align features across scales using the least common multiple scaling, then enhance the interaction and integration of data across scales through a bidirectional processing strategy and modified Mamba architectures, specifically ST-Mamba and Cross-Mamba. Evaluated on six real-world datasets, our study demonstrates the effectiveness of the proposed framework in handling multi-scale data completion challenges, particularly when dealing with non-aligned compositional relationships.

Wide Area Network (WAN) management faces significant complexity in characterizing heterogeneous traffic from diverse services, which complicates unified feature extraction for operational tasks like anomaly detection. Key challenges persist in difficulty of creating generic feature sets across diverse traffic patterns and insufficient historical data for machine learning (ML) generalization. Meta-learning algorithms can be applied to learn dynamic feature sets to improve model generalizability on a limited number of data records. However, prior research focuses on algorithm design, with a lack of study on its application to in-band network monitoring. This work proposes a workflow to adaptively collect new feature sets and promptly learn from them. An adaptive in-band feature selection and extraction method is proposed for programmable switch. Meta learning algorithm is introduced for prompt decision in controller based on few-shot records. Evaluation results have shown that it outperforms prior methods in accuracy and inference time on public datasets.

Neural Architecture Search (NAS) can discover the optimal neural network architecture within a given SuperNet through automated search, which can improve model performance and reduce computational overhead on resource-constrained devices. Due to the vast SuperNet requiring substantial computational resources for training and evaluation, the search process is costly and difficult to apply directly on End Devices (EDs) with limited computational resources. Moreover, existing methods utilize zero-cost proxies to reduce computational costs in NAS, but overlook limited computational resources on EDs and waste a large amount of computational resources on the cloud server. Deploying NAS on the cloud server can effectively address this issue. The cloud server is used to search for the optimal Subnet, and EDs only need to train the Subnet based on local data. To this end, we propose a nonlinear aggregation-based Neural Architecture Search method based on Feature and Gradient zero-cost proxies (FG-NAS). Specifically, EDs upload local data characteristics to the cloud server, and then the cloud server uses FG-NAS to obtain an optimal SubNet model from the SuperNet based on the uploaded data characteristics. Finally, the cloud server sends the optimal SubNet to EDs, which can reduce the computational burden on EDs. Furthermore, FG-NAS evaluates the accuracy of neural architectures by considering both feature proxies during forward propagation and gradient proxies during backward propagation. Experiments on three datasets demonstrate that compared to current mainstream zero-cost proxy methods, FG-NAS can improve evaluation accuracy by an average of 1.04% and reduce single-network evaluation time by up to 2.45%.

Intrusion detection systems (IDS) primarily rely on signature-based approaches, which can fail to detect novel or sophisticated attacks. This paper addresses the underutilized potential of leveraging a multi-modal approach that combines packet capture (PCAP) and log data for anomaly detection. To enhance detection capabilities, we propose an interpretable hybrid neural network architecture, TransIDS, that integrates a packet-based transformer with an efficient transformer-based language model for log messages. The proposed framework extracts semantic vectors from raw log messages and concatenates them with packet embeddings. An attention-based classification model then detects anomalies by determining the importance of each log message and packet for the neural network’s decision. By fusing spatial features from PCAP data with temporal features from log data, TransIDS utilizes this multi-modal data fusion to identify anomalies that might be missed by conventional systems. This approach not only leverages the strengths of two distinct transformer-based architectures but also provides a more comprehensive analysis of network traffic, leading to more effective detection of previously undetected attacks and strengthening overall network security. We use a real testbed for our experiments to validate the effectiveness of our proposed approach.

In UAV applications, dense haze severely obscures small ground-level objects, hindering the recovery of fine details. Existing visible-only dehazing methods struggle with such dense occlusions, while infrared imaging lacks color and fine texture information. To address these limitations, we propose the Haze Distribution-aware Cross-modal Fusion Network (HDCFN). HDCFN features two key components: (i) an infrared-guided multiscale feature enhancement framework that integrates haze-resistant structural cues from infrared modality with visible features across coarse to fine, improving the recovery of small objects, and (ii) a haze distribution-aware cross-modal fusion module that adaptively prioritizes relevant information from each modality according to haze density. This framework effectively combines the complementary strengths of visible and infrared imaging for dense haze removal. Extensive experiments on multiple public datasets show that HDCFN outperforms state-of-the-art dehazing and fusion methods, yielding higher-quality and more detailed images.

Small language models (SLMs) support efficient deployments on resource-constrained edge devices, but their limited capacity compromises inference performance. Retrieval-augmented generation (RAG) is a promising solution to enhance model performance by integrating external databases, without requiring intensive on-device model retraining. However, large-scale public databases and user-specific private contextual documents are typically located on the cloud and the device, respectively, while existing RAG implementations are primarily centralized. To bridge this gap, we propose DRAGON, a distributed RAG framework to enhance on-device SLMs through both general and personal knowledge without the risk of leaking document privacy. Specifically, DRAGON decomposes multi-document RAG into multiple parallel token generation processes performed independently and locally on the cloud and the device, and employs a newly designed Speculative Aggregation, a dual-side speculative algorithm to avoid frequent output synchronization between the cloud and device. A new scheduling algorithm is further introduced to identify the optimal aggregation side based on real-time network conditions. Evaluations on real-world hardware testbed demonstrate a significant performance improvement of DRAGON—up to 1.9X greater gains over standalone SLM compared to the centralized RAG, substantial reduction in per-token latency, and negligible Time to First Token (TTFT) overhead.

Inspired by the impressive reasoning capabilities demonstrated by reinforcement learning approaches like DeepSeek-R1, recent emerging research has begun exploring the use of reinforcement learning (RL) to enhance vision-language models (VLMs) for multimodal reasoning tasks. However, most existing multimodal reinforcement learning approaches remain limited to spatial reasoning within single-image contexts, yet still struggle to generalize to more complex and real-world scenarios involving multi-image positional reasoning, where understanding the relationships across images is crucial. To address this challenge, we propose a general reinforcement learning approach PeRL tailored for interleaved multimodal tasks, and a multi-stage strategy designed to enhance the exploration-exploitation trade-off, thereby improving learning efficiency and task performance. Specifically, we introduce permutation of image sequences to simulate varied positional relationships to explore more spatial and positional diversity. Furthermore, we design a rollout filtering mechanism for resampling to focus on trajectories that contribute most to learning optimal behaviors to exploit learned policies effectively. We evaluate our model on 5 widely-used multi-image benchmarks and 3 single-image benchmarks. Our experiments confirm that PeRL trained model consistently surpasses R1-related and interleaved VLM baselines by a large margin, achieving state-of-the-art performance on multi-image benchmarks, while preserving comparable performance on single-image tasks.

This paper investigates the illusionary difficulty novice programmers face, where students struggle with new tasks despite feeling well-prepared. The study explicitly shows that traditional assessments, such as self-reports, often fail to identify the misconceptions underlying this challenge. Using Veenman's metacognitive cueing with scaffolded self-check questions, we help students -- without introducing additional materials -- to refocus from their current understanding to uncovering unconscious knowledge gaps. This approach encourages students to step out of their comfort zones, engage in deeper self-exploration, persist through real difficulties, accommodate new concepts, and resolve misconceptions in their grasp. Pilot experiments reveal that while students initially overestimate their understanding, applying this cueing approach during test preparation significantly improves their actual test performance, as opposed to cumulative performance that relies on feedback and reinforcement through practices. The results underscore the importance of addressing misconceptions in problem-solving training using non-identical cues (as opposed to code evaluation training, which follows a predetermined solution path) to foster long-term self-efficacy, even if it temporarily reduces self-esteem.

Monitoring Key Performance Indicator (KPI) trends in a timely manner enables early detection of performance degradation, which is critical to maintaining cloud service reliability. However, incipient anomalies that subtly precede KPI degradation are notoriously difficult to detect due to interference from noise and the high-dimensional, correlated nature of monitoring data. In modern production environments, KPIs are recorded as massive multivariate time series (MTS), where intricate temporal and inter-metric dependencies further obscure early signals of system instability. Unfortunately, existing fault detection methods either rely on oversimplified statistical assumptions that ignore these dependencies or depend on largescale supervised training, which is rarely feasible in dynamic, label-scarce settings. We introduce Heimdallr, an unsupervised detection framework tailored to uncover early-stage anomalies that causally affect KPIs. The core design of Heimdallr is a KPI-oriented monitoring and attribution mechanism that models and partitions the latent space based on KPI behavior. This design enables not only the early detection of anomalies but also causal attribution to specific KPI shifts, thereby enhancing system observability and resilience. Heimdallr is built upon two key innovations: Dynamic-inner Related Component Analysis (DiRCA), a latent structure modeling technique that captures dynamic temporal dependencies across metrics, enabling interpretable representations of underlying system behavior; KPI Feature Ensemble Monitoring Network (KFEMNet), a three-layer hierarchical architecture that, following DiRCA's decomposition, extracts fine-grained deviations and detects incipient anomalies with high sensitivity. Extensive experiments on both synthetic and real-world datasets convincingly demonstrate that Heimdallr consistently outperforms existing state-of-the-art methods, achieving higher early detection accuracy and lower false alarm rates, while maintaining low overhead and high interpretability suitable for production deployment.

A Link Flooding (LFA) attack is a special type of Denial-of-Service (DoS) attack in which the attacker sends out a huge number of requests to exhaust the capacity of a link on the path the traffic comes to a server. As a result, user traffic cannot reach the server. As a result, DoS and degradation of Quality-of-Service (QoS) occur. Because the attack traffic does not go to the victim, protecting the legitimate traffic alone is hard for the victim. The victim can protect its legitimate traffic by using a special type of router called filter router (FR). An FR can receive server filters and apply them to block a link incident to it. An FR probabilistically appends its own IP address to packets it forwards, and the victim uses that information to discover the traffic topology. By analyzing traffic rates and paths, the victim identifies some links that may be congested. The victim needs to select some of these possible congested links (PCLs) and send a filter to the corresponding FR so that legitimate traffic avoids congested paths. In this paper, we formulate two optimization problems for blocking the least number of PCLs so that the legitimate traffic goes through a non-congested path. We consider the scenario where every user has at least one non-congested shortest path in the first problem. We extend the first problem to a scenario where there are some users whose shortest paths are all congested. We transform the original problem to the vertex separation problem to find the links to block. We use a custom-built Java multi-threaded simulator and conduct extensive simulations to support our solutions.

Traditional single-machine Network Intrusion Detection Systems (NIDS) are increasingly challenged by rapid network traffic growth and the complexities of advanced neural network methodologies. To address these issues, we propose an Enhanced Meta-IDS framework inspired by meta-computing principles, enabling dynamic resource allocation for optimized NIDS performance. Our hierarchical architecture employs a three-stage approach with iterative feedback mechanisms. We leverage these intervals in real-world scenarios with intermittent data batches to enhance our models. Outputs from the third stage provide labeled samples back to the first and second stages, allowing retraining and fine-tuning based on the most recent results without incurring additional latency. By dynamically adjusting model parameters and decision boundaries, our system optimizes responses to real-time data, effectively balancing computational efficiency and detection accuracy. By ensuring that only the most suspicious data points undergo intensive analysis, our multi-stage framework optimizes computational resource usage. Experiments on benchmark datasets demonstrate that our Enhanced Meta-IDS improves detection accuracy and reduces computational load or CPU time, ensuring robust performance in high-traffic environments. This adaptable approach offers an effective solution to modern network security challenges.

Visual information is first encoded into spike trains by retinal ganglion cells (RGCs), forming the foundation of biological vision. Decoding the semantic and perceptual content embedded in these spikes remains a key challenge in neuroscience. We introduce a novel framework that leverages recent advances in artificial intelligence to interpret spike-based visual signals. Our framework consists of spike-driven multimodal large language model (MLLM) and a spike-driven ControlNet to emulate two core functions of the visual cortex: semantic understanding and image reconstruction. By bridging biological and computational vision, our framework enables effective decoding of spike signals generated by RGCs. We further validate the model on the spike-driven MLLM on the MSCOCO test set. The experiment results demonstrate that our framework not only enhances interpretability and usability of spike-driven outputs but also holds promise as a tool to support neuroscience research and aid visually impaired individuals. Participant feedback highlights its potential to facilitate future advances in neural decoding and vision restoration technologies.

The lifetime of a sensor network is generally longer than that of a single sensor node, so to ensure the connectivity of the network, new nodes need to be deployed into the network in multiphase. Such networks are called multiphase sensor networks. In sensor networks, establishing a pairwise key between two neighboring nodes is a fundamental security requirement. However, for sensor nodes with strictly limited capabilities, there are great challenges in establishing a pairwise key between two neighboring sensor nodes deployed at different phases. In order to solve this problem, some scholars have proposed some solutions based on backward key-chains and forward key-chains, however, how to further improve the networks’ resilience against node capture attacks still needs further research; and some scholars have proposed some solutions based on key pools of each phase are independent, but how to improve the probability of establishing a pairwise key between two neighboring nodes deployed at different phases is still a very challenging problem. In this article, we propose a new pairwise key scheme based on deployment knowledge. Theoretical analysis and simulation show that the proposed model can achieve high probability of establishing a pairwise key between two neighboring nodes and good resilience against node capture attacks by setting appropriate parameters.

Sensor networks are often deployed outdoors and are more susceptible to various attacks. In order to protect communication between nodes, scholars have proposed key management schemes. The most popular among them is the key predistribution scheme. This scheme has the following contradiction: in order to improve security, the number of keys predistributed to nodes should be minimized as much as possible. However, as the number of predistribution keys decreases, the network may no longer be securely connected, resulting in wastage of nodes. In this letter, we propose a new key establishment method to address this issue. Analysis and simulation show that the proposed scheme can reduce the number of predistribution keys in the original schemes while ensuring secure network connectivity.

In real-world Industrial Internet of Things (IIoT) scenarios, due to the limited storage capacity of IIoT devices, fresh data continuously received by diverse devices will overwrite the outdated data and change the local data distribution. However, state-of-the-art studies have demonstrated that federated learning tends to focus on training with fresh data, and the latest global model may forget the historical update directions (i.e., catastrophic forgetting). This issue can significantly degrade the global model accuracy. Existing methods primarily focus on integrating outdated data characteristics into fresh data but overlook the large parameter update gap between global and local models during global aggregation. This gap can cause the global model updates to deviate from the optimal direction. To this end, we propose a federated adaptive weighted aggregation method based on model consistency (FedAWAC). Specifically, FedAWAC measures the model consistency on devices and dynamically adjusts the aggregation weights of each local model, thereby guiding the global model toward optimal updates. Furthermore, FedAWAC integrates$\mathcal {M}$historical global models most correlated to the latest global model on the cloud server to overcome catastrophic forgetting. Experiments on four different datasets (nonidentically and independently distributed settings) indicate that compared to five baselines, FedAWAC can improve global model accuracy by an average of 1.86%, reduce the forgetting rate by an average of 3.91%, and save average memory usage by up to 2.57 GB.

Cloud-centric deployment of large models brings privacy concerns and communication burdens to IoT devices. Federated Learning (FL) can help numerous IoT devices collaboratively train a large model in a distributed manner to provide intelligent services and applications (AIoT). However, due to the limited storage capacity of IoT devices, fresh data collected by IoT devices often overwrites outdated data and establishes Non-IID data distributions. Moreover, state-of-the-art studies have indicated that FL tends to use fresh data for model training. This causes the global model to forget outdated data’s characteristics (i.e., catastrophic forgetting). Some methods utilize Knowledge Distillation (KD) to extract and integrate characteristics from both fresh data and outdated data. However, existing KD-based methods use fixed distillation temperatures for different IoT devices, which overlooks that fixed distillation temperatures cannot match the data distributions on different IoT devices and may degrade global model accuracy. To this end, we propose a Federated Dynamic Decoupled Distillation method based on Logits distribution (Fed3DL). Specifically, Fed3DL utilizes decoupled distillation to mitigate catastrophic forgetting. To dynamically adjust the distillation temperature of each IoT device, Fed3DL novelly builds an adaptive temperature-aware mechanism based on the Logits distribution. Furthermore, Fed3DL introduces a regularization term into the local distillation loss to improve global model accuracy. Experiments on four datasets show that Fed3DL can improve the global model accuracy by an average of 3.19%, reduce the forgetting rate by an average of 4.46%, and achieve the lowest inter-class accuracy disparity.

In multi-phase sensor networks, each deployment is called a phase of networks, and how to improve the secure connectivity between nodes deployed at different phases is a key issue. But this is not an easy task for resource-constrained sensor nodes. Recently, some scholars have proposed some methods where nodes deployed at the same phase can establish shared keys directly, which offers perfect resilience against destruction. Additionally, they introduced an online key update method to establish shared keys for nodes deployed at different phases. However, the resilience of this scheme decreases rapidly as the number of newly deployed nodes captured during the key establishment phase increases. In this paper, we propose a T-type key pre-distribution scheme based on deployment knowledge, and also introduce a novel method for path key establishment. In this scheme, the shared keys established directly between nodes exhibit perfect resilience. Furthermore, this scheme does not require the assumption of a relatively secure time period during the key establishment phase. In this proposed scheme, by setting appropriate parameters, theoretical analysis and simulation show that the probability of two neighboring nodes establishing a shared key is about 1, and the resilience against node capture attacks is excellent.

The secure container that hosts a single container in a micro virtual machine (VM) is now used in serverless computing, as the containers are isolated through the microVMs. There are high demands on the high-density container deployment and high-concurrency container startup to improve both the resource utilization and user experience, as user functions are fine-grained in serverless platforms. Our investigation shows that the entire software stacks, containing the cgroups in the host operating system, the guest operating system, and the containerrootfsfor the function workload, together result in low deployment density and slow startup performance at high-concurrency.We propose a lightweight and holistic-scalable secure container runtime, named RunD-V, to resolve above problems in serverless computing. RunD-V proposes a guest-to-host runtime template for microVM scaling-out, and CR-bind feature in guest kernel for microVM scaling-up. Using guest-to-host runtime template, over 200 secure containers can be launched within 1son a node equipped with 104 vCPUs. It also enables more than 2,500 secure containers to be deployed on a node with 384GB of memory. The vertical scaling mechanism CR-bind further enhances both startup concurrency and deployment density.

We are now embracing an era of edge computing and artificial intelligence, and the combination of the two has spawned a new field of research called edge intelligence. Massive amounts of data is generated at the edge of network, which relies on artificial intelligence to realize its potential. Meanwhile, artificial intelligence is able to flourish when processing diverse edge data. However, the computation and storage resources of edge servers are not unlimited. For some large-scale intelligent applications, it is difficult to meet their service quality requirements by directly offloading the entire application to a nearby server for processing. Due to the heterogeneity of server resources in edge environments, how to balance the workload among edge servers to provide better services also becomes complicated. The goal of this paper is to minimize the total cost of offloading large-scale applications consisting of many dependent tasks in an edge system. We formulate the Dependent task Offloading with Service Caching (DOSC) problem, which is proved to be NP-hard. A dynamic planning-based algorithm is introduced to solve fixed-DOSC, in which some services are pre-configured on the edge server, and other services can not be downloaded from the remote cloud. We also present a theoretical analysis on the performance guarantee of the dynamic planning-based algorithm. Then, we propose a near-optimal algorithm using the Gibbs sampling to solve the general DOSC problem. Testbed experiments and trace-driven simulations are conducted to verify the performance of our algorithm. Our algorithm, shown to be the most effective in terms of cost, considers both service caching and task dependencies when task offloading in comparison to other baseline algorithms.

Federated Learning with Cloud-Edge Collaboration (FL-CEC) has emerged as a cutting-edge paradigm in distributed learning. Efficient resource investment incentive mechanisms are crucial to encouraging clients in FL-CEC to contribute the necessary data and computational resources for training. However, existing studies are inadequate in meeting the incentive design requirements under multi-level information-sharing scenarios. Moreover, current works often rely on specific functional relationships between resource investment and global model accuracy. To bridge these gaps, this paper investigates the incentive problem for data and computational resource investment under multi-level information-sharing levels. We design a resource investment incentive mechanism based on a weighted potential game without depending on any specific functional relationship between data investment and model accuracy. Furthermore, we propose four algorithms to solve resource investment strategies for different levels of information sharing. The complexity and convergence rates of the proposed algorithms are thoroughly analyzed. Finally, we construct a simulation incentive platform on the Aliyun. Extensive evaluations demonstrate that the proposed scheme effectively enhances social welfare, and improves collaborative training accuracy and efficiency.

Previous state-of-the-art studies have demonstrated that adversaries can access sensitive user data by membership inference attacks (MIAs) in Federated Learning (FL). Introducing differential privacy (DP) into the FL framework is an effective way to enhance the privacy of FL. Nevertheless, in differentially private federated learning (DP-FL), local gradients become excessively sparse in certain training rounds. Especially when training with low privacy budgets, there is a risk of introducing excessive noise into clients’ gradients. This issue can lead to a significant degradation in the accuracy of the global model. Thus, how to balance the user's privacy and global model accuracy becomes a challenge in DP-FL. To this end, we propose an approach, known as differential privacy federated aggregation, based on significant gradient protection (DP-FedASGP). DP-FedASGP can mitigate excessive noises by protecting significant gradients and accelerate the convergence of the global model by calculating dynamic aggregation weights for gradients. Experimental results show that DP-FedASGP achieves comparable privacy protection effects to DP-FedAvg and cpSGD (communication-private SGD based on gradient quantization) but outperforms DP-FedSNLC (sparse noise based on clipping losses and privacy budget costs) and FedSMP (sparsified model perturbation). Furthermore, the average global test accuracy of DP-FedASGP across four datasets and three models is about$2.62$%,$4.71$%,$0.45$%, and$0.19$% higher than the above methods, respectively. These improvements indicate that DP-FedASGP is a promising approach for balancing the privacy and accuracy of DP-FL.

The end-edge-cloud-based Social Internet of Things (SIoT) faces increasing threats from malware. To address these challenges, we propose an explainable novel malware defense framework that integrates Markov games with multi-agent deep reinforcement learning under an end-edge-cloud-based SIoT collaborative architecture. The framework models the interactions between malicious SIoT nodes and edge devices as a multi-agent game problem, incorporating multi-layer defense mechanisms to achieve precise descriptions of attack-defense behaviors. By combining Markov games with the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm, we develop the Robust Multiagent Actor-Critic (RMAAC) algorithm, which enables adaptive strategy optimization. To enhance system interpretability, we introduce SHapley Additive exPlanations (SHAP) value analysis into the defense decision-making process, providing transparent insights into feature contributions and decision rationales. Extensive experimental results demonstrate that the proposed RMAAC algorithm significantly outperforms existing methods, including MADDPG and Minimax Multi-Agent Deep Deterministic Policy Gradient (M3DDPG), in multiple performance metrics such as episode reward, hacking success rate, and cumulative defense number. Through systematic parameter optimization, including batch size and agent interaction speed, the framework provides an effective and sustainable solution for explainable malware defense in SIoT environments.

Federated learning (FL) is a distributed learning paradigm that enables large-scale IoT devices to collaboratively train a shared model while preserving the privacy of local data. To avoid the single-point-of-failure of the conventional parameter server architecture, the study concentrates on the decentralized FL (DFL) paradigm building on the device-to-device communication network. However, existing DFL frameworks encounter challenges related to resource limitations, privacy protection, and data heterogeneity. To overcome these challenges, the study proposes and implements DF$^{2}$-MPC in industrial IoT, an efficient DFL framework with personalized model pruning and adaptive communication. Specifically, a personalized pruning ratio determination approach is designed by exploiting the model pruning technique. This approach enables all devices to flexibly determine pruning ratios by themselves, thereby achieving both communication savings and privacy protection. Then, this study designs an adaptive neighbor selection scheme, which can enhance model performance and foster model consensus under resource constraints. In addition, the study theoretically proves the convergence performance of DF$^{2}$-MPC. Finally, extensive simulations on three real-world traces are conducted to corroborate the superiority of DF$^{2}$-MPC, demonstrating that the method can improve communication efficiency with satisfactory model accuracy and convergence performance.

In online social networks (OSNs), people usually join groups for communication. Information diffusion often occurs with some cost, either between individuals or within/among groups; and different opinions may compete with each other. The groups can make decisions based on the majority of the group members. This type of group consensus is common in group activities. However, existing research on maximization of competitive influence often neglects the effects of group consensus. To this end, we introduce the process of group consensus reaching in influence maximization and propose a novel Group consensus-based Competitive Linear Threshold (GCLT) propagation model; then we study the Budgeted Competitive Influence Maximization (BCIM) problem under the GCLT model. We reveal that the problem is NP-hard, and the objective function is proven to be neither submodular nor supermodular. To this end, we construct an equivalent Group consensus-based Competitive Live-Edge (GCLE) model of GCLT by sampling method. Based on GCLE, develop two submodular functions of the upper and lower bounds. Then, we propose the SBG algorithm by applying the Sandwich Approximation framework for the BCIM problem under the GCLT model. In SBG , we provide an approximate solution to the lower bound and the upper bound by the proposed OPIM-B algorithm. Then, we select the seed set of solutions that achieves the best influence spread in Monte Carlo simulations. We also propose two strategies to optimize SBG . The experiments on six real social network datasets verify the effectiveness and scalability of our method and validate the impact of group consensus on the competitive influence dissemination process, as well as the importance of considering the process of reaching group consensus.

In recommender systems, it is frequently presumed that missing ratings adhere to a missing at random (MAR) mechanism, implying the absence of ratings is independent of their potential values. However, this assumption fails to hold in real-world scenarios, where users are inclined to rate items they either strongly favor or disfavor, introducing a missing not at random (MNAR) scenario. To tackle this issue, prior researchers have utilized explicit MAR feedbacks to infer the propensities of unobserved, implicit MNAR feedbacks. Nonetheless, acquiring explicit MAR feedbacks is resource-intensive and time-consuming and may not reflect users’ true preferences. Furthermore, most methods have only been tested on synthetic or small-scale datasets, thus their applicability and effectiveness in real-world settings without MAR feedbacks remain unclear. Along these lines, we aim to predict MNAR ratings without MAR prior propensities by exploring the consistency between MAR and MNAR feedbacks and narrowing the gap between them. From the empirical study and preliminary experiment, we hypothesize thatuser preferencescan be treated as the common prior propensity for both MAR and MNAR generative processes. In this way, we extend this hypothesis to a more general MNAR scenario: user preferences learned from MNAR can partially substitute for the prior propensities derived from MAR feedbacks for MNAR recommendation tasks. To validate our hypothesis and approach, we develop a lightweight iterative probabilistic matrix factorization framework (lightIPMF) as a practical method of our methodology, utilizing user preferences extracted from MNAR, not MAR, to estimate MNAR feedbacks. Finally, the experimental results show that modeling user preferences can effectively improve MNAR feedback estimation without MAR feedback, and our proposed lightIPMF outperforms the state-of-the-art MNAR methods in predicting MNAR feedbacks.

Federated learning participants have two identities: model trainers and model users. As model users, participants care most about the performance of the final model on their own distributions, which is called Personal Model Performance (PMP). This makes training a single global model to accommodate all participants impractical because the data distributions of participants are heterogeneous. As model trainers, due to high training costs, participants are reluctant to contribute models if incentives are not enough. With the combination of the above two reasons, we propose a dual-identity double auction as an incentive mechanism in personalized federated learning, allowing directional selection between model users and model trainers, both of which are served by FL participants. Within the double auction framework, we devise a reinforcement learning-based model selection method. This method selects a set of models for each buyer to bid on. The bought models are aggregated to be a personalized model to achieve higher PMP. Additionally, we implement a transaction partition-based approach for determining clearing prices and winning pairs. We address the challenge of the unavailability of private yet essential data distribution information, the coupled influence of model selection and auction results on PMP, and more utility improvement ways of multi-demand dual-identity participants. Finally, our double auction optimizes the PMP of all participants and ensures the truthfulness of multi-demand dual-identity participants, which is harder compared with single-demand single-identity participants.

Massively deployed cameras form a tightly connected network which generates video streams continuously. Benefiting from advances in computer vision, automated real-time analytics of video streams can be of practical value in various scenarios. As cameras become more dense, cross-camera video analytics has emerged. Combining video contents from multiple cameras for analytics is certainly more promising than single-camera analytics, which can realize cross-camera pedestrian tracking and cross-camera complex behavior recognition. Some works focused on optimization of cross-camera video analytic applications, but most of them ignore specific network situation between cameras and edge servers. Furthermore, most of them ignore the super resolution technique, which is proven to be a source of efficiency. In this paper, we first verify the potential gain of super resolution on cross-camera video analytic tasks. Then, we design and implement a cross-camera real-time video streaming analytic system,${\mathsf {Scrava}}$, which leverages super resolution to augment low-resolution videos and simultaneously reduce bandwidth consumption.${\mathsf {Scrava}}$enables real-time cross-camera video analytics and enhances video segments with the SR module under poor network conditions. We take cross-camera pedestrian tracking as an example, and experimentally verifies the effectiveness of super resolution on real-time cross-camera video analytics. Compared with using low-resolution video segments,${\mathsf {Scrava}}$can improve the F1 score by 47.16%, verifying the feasibility of exploiting super resolution to improve the performance of real-time cross-camera video analytic systems.

Mobile CrowdSensing (MCS) has emerged as a popular paradigm to engage mobile users in collaborative sensing tasks. However, its performance is hindered by its limited spatiotemporal range and the cost of data collection. An effective strategy is to integrate Sparse MCS with data completion, allowing for unsensed data inference. However, when confronted with situations where sensed data is excessively sparse, data inference results may be unsatisfactory due to several challenges including: 1) uneven data distribution, 2) complex spatiotemporal correlation, and 3) the presence of inference noise. To address these challenges, we propose a model named Spatiotemporal Evolutionary Inference (STEI) that achieves accurate inference of unsensed data in Sparse MCS. Specifically, we complete the unsensed data by uncovering strong local correlations in the data and gradually evolving those correlations to the global situation. In each evolution step, we thoroughly consider the impact of spatiotemporal consistency and difference. To minimize the interference of noise during the evolution process, we design an adaptive coefficient to enhance the dependence on sensed data. Finally, to validate the effectiveness of STEI, we conduct extensive qualitative and quantitative experiments using three popular datasets. The experimental results demonstrate that our approach excels in accurately inferring data, particularly in situations where the distribution of data is notably uneven.

With the development of large-scale artificial intelligence services, edge devices are becoming essential providers of data and computing power. However, these edge devices are not immune to malicious attacks. Federated learning (FL), while protecting privacy of decentralized data through secure aggregation, struggles to trace adversaries and lacks optimization for heterogeneity. We discover that FL augmented with Differentiable Architecture Search (DARTS) can improve resilience against backdoor attacks while compatible with secure aggregation. Based on this, we propose a federated neural architecture search (NAS) framwork named SLNAS. The architecture of SLNAS is built on three pivotal components: a server-side search space generation method that employs an evolutionary algorithm with dual encodings, a federated NAS process based on DARTS, and client-side architecture tuning that utilizes Gumbel softmax combined with knowledge distillation. To validate robustness, we adapt a framework that includes backdoor attacks based on trigger optimization, data poisoning, and model poisoning, targeting both model weights and architecture parameters. Extensive experiments demonstrate that SLNAS not only effectively counters advanced backdoor attacks but also handles heterogeneity, outperforming defense baselines across a wide range of backdoor attack scenarios.

Mobile Opportunistic Networks (MONs) often experience frequent interruptions in end-to-end connections, which increases the likelihood of message loss during delivery and makes users more susceptible to various cyber attacks. However, most currently proposed anonymous routing protocols are primarily designed for networks with stable connections, making it challenging to protect user identities in MONs. To address these challenges, we propose FLAG-POR (Forward Legal Anonymous Group Pairing-Onion Routing), a novel anonymous routing protocol specifically tailored to enhance message delivery anonymity and security in MONs. Specifically, we abstract the mobile opportunistic network as a contact graph. By introducing the concept of “groups” into the pairing-onion routing protocol, which encrypts messages and relay nodes layer by layer, we develop a novel group-based pairing-onion routing protocol. This protocol ensures message confidentiality and relay node anonymity, while also improving message forwarding rates, as any node within a group can potentially act as a relay. To ensure message authenticity, we employ the efficient SM2 signing algorithm to generate signatures for the message source. Furthermore, by incorporating parameters such as the public key validity period and master key validity period into the group pairing-onion routing protocol, we achieve forward security in message delivery. We conduct a thorough theoretical analysis of the protocol’s security and performance. The experimental results demonstrate that our FLAG-POR protocol outperforms baseline anonymous protocols in terms of delivery success rate, traceability rate, path anonymity, and node anonymity. Additionally, the FLAG-POR scheme effectively resists three potential threats to the routing system: collusion attack threat, node identification threat, and path identification threat, in any situation.

Edge video analytics typically rely on conventional encoding standards to transmit device visual data for server-side inference. Unfortunately, general-purpose compression solutions retain unnecessary visual data that does not contribute to accuracy, resulting in significant latency throughout Video Analytics Pipeline (VAP). While previous approaches have made partial progress, they cannot systematically eliminate VAP redundancy due to uncoordinated subsystem-level optimization. Achieving complete redundancy elimination presents a major challenge, as a lack of spatio-temporal coordination risks offsetting latency gains with computational overhead (associated with redundancy elimination).Crucioovercomes these limitations with an end-to-edge framework that integrates temporally adaptive frame filtering and coordinated video compression. It leverages redesigned asymmetric autoencoders to synchronize inter-frame temporal compression with intra-frame spatial feature extraction. Additionally,Crucioemploys a one-pass decoding mechanism for encoded critical frames and dynamically adjusts batching scales to minimize latency. Empirical results demonstrateCrucio's superiority, outperforming existing solutions (e.g., DDS, Reducto, and STAC) by over a 31% reduction in end-to-end latency at 0.9 accuracy thresholds.

In the context of mobile edge computing, achieving dynamic service caching and updating to guarantee the QoS of users and reduce system costs is a challenging problem. However, existing research still has certain deficiencies in considering the dynamic behavior of users and the limited storage resources of edge servers. To address this problem, this paper investigates optimizing the service caching and updating problem within multi-stage and proposes a novel framework with three proposed strategies for the different stages to jointly optimize the delay and cost. At the initial service caching stage, we propose a basic caching strategy based on dynamic programming for the single-area scenario, taking into account the constraint of limited memory resources. To improve the caching strategy, we extend our consideration to the multiple-area scenario and design an improved algorithm based on tabu search. Given the dynamic behavior of users, we formulate the joint optimization problem as a Markov Decision Process (MDP) and design a service extension strategy based on reinforcement learning at the service updating decision-making stage and a replacement strategy taking both the distribution of service replications and service access frequency into account at the service updating replacement stage to guarantee the QoS of users. We effectively tackle the challenges arising from the dynamic behavior of users and limited storage resources. Through extensive comparative experiments, our approach outperforms traditional strategies by significantly reducing user latency and system cost.

The serendipity recommendation tries to burst the filter bubble while still meeting user interests. However, serendipity itself has not been well understood in the recommendation system. Thus, factor investigation in recommendation serendipity has attracted much attention, for which two challenges hinder follow-up research: (1) Ambiguity of factors . Different works exploit different factors, and the meanings of factors are inconsistent in various works. (2) Lack of complete impact validation . The importance of these factors in different domains is not yet fully understood. The common approach of user surveys costs much, but the results are usually less objective and limited in quantity. To this end, we strive to comprehensively identify and clarify serendipity factors and explore objective data-driven approaches to validate factor impacts in large-scale cross-domain scenarios. We first conduct a comprehensive literature review to identify all possible factors, from which we find that some factors are being used indistinguishably. To address this issue, we propose two principles of meaning coverage and factor independence to clarify and disentangle serendipity factors. Next, we propose a general experimental framework to explore the impacts of factors. Then, we implement one such framework and run experiments on nine representative datasets to study factor importance on serendipity. We also propose a quantitative method to measure the degree of disentanglement of factors and to test the effects of factor combinations. We gain several useful findings: (1) relevance , diversity , and random are critical factors affecting serendipity; (2) domain features affect factor importance and can guide serendipity recommendation; (3) the disentanglement quantification method benefits the understanding of serendipity and the combination of factors. To our knowledge, this is the first work to comprehensively investigate serendipity factors and experimentally compare their impacts in an objective data-driven approach.

Running deep neural networks (DNNs) on large-scale videos from widely distributed cameras presents two significant challenges. Firstly, video quality for analytical purposes is severely impacted by the camera deployment environment, which is termed Pixel Recession in this paper. Secondly, low-latency video streaming from the source camera to edge servers is greatly hindered by the rapid expansion of video traffic. Despite numerous efforts such as enhancing the video structure, uneven encoding, and filtering frames captured on camera, these methods have proven insufficient to address the challenges at hand. We propose Spliceosome, a novel video analytics system that effectively overcomes the pixel recession and streaming bottlenecks. In brief, Spliceosome 1) recovers from pixel recession by adaptive video knobs (i.e., brightness and contrast) tuning in ARP (anchor region proposal) granularity, and 2) lowers the transmission volume by video thinning, which uses only single-channel information for video encoding. We implemented Spliceosome using only commercial off-the-shelf hardware. Our experimental results demonstrate that Spliceosome outperforms other alternative designs by 4.71-14.47%, 40.94-58.71%, and 14.28% in detection accuracy, end-to-end delay, and efficiency of DNNs inference, respectively.

In cloud data centers, the exponential growth of data places increasing demands on computing, storage, and network resources, especially in multi-tenant environments. While this growth is crucial for ensuring Quality of Service (QoS), it also introduces challenges such as fluctuating resource requirements and static container configurations, which can lead to resource underutilization and high energy consumption. This article addresses online resource provisioning and efficient scheduling for multi-tenant environments, aiming to minimize energy consumption while balancing elasticity and QoS requirements. To address this, we propose a novel optimization framework that reformulates the resource provisioning problem into a more manageable form. By reducing the original multi-constraint optimization to a container placement problem, we apply the interior-point barrier method to simplify the optimization, integrating constraints directly into the objective function for efficient computation. We also introduce elasticity as a key parameter to balance energy consumption with autonomous resource scaling, ensuring that resource consolidation does not compromise system flexibility. The proposed Energy-Efficient and Elastic Resource Provisioning (EEP) framework comprises three main modules: a distributed resource management module that employs vertical partitioning and dynamic leader election for adaptive resource allocation; a prediction module using$\omega$-step prediction for accurate resource demand forecasting; and an elastic scheduling module that dynamically adjusts to tenant scaling needs, optimizing resource allocation and minimizing energy consumption. Extensive experiments across diverse cloud scenarios demonstrate that the EEP framework significantly improves energy efficiency and resource utilization compared to established baselines, supporting sustainable cloud management practices.

The rise of software-defined networking (SDN) has greatly enhanced the flexibility and programmability of networks, but it has also brought new security challenges while achieving more efficient multipath, more flexible load balancing, and more secure authentication. To address this problem, we propose an innovative solution: the load-balanced heterogeneous multipath authenticated routing selection (LBHMARS) framework. This framework implements load balancing of multipath routing, as well as authentication and location of compromised switches in SDN networks, while ensuring the security of traffic entering the system. LBHMARS establishes a secure and efficient routing mechanism in SDN using the IPv6 protocol. We combine strong encryption and signature schemes with a novel path load balancing algorithm to ensure traffic balance and data integrity among multiple network paths. Finally, we introduce a mechanism to locate and isolate malicious switches, further enhancing network security. Compared with existing solutions, our extensive simulations show that LBHMARS has superior performance in mitigating attacks and maintaining high network throughput. In addition, LBHMARS also performs well in terms of latency and low packet loss.

In edge computing, end devices (EDs) containerize tasks with the necessary resources and offload subsets to a nearby high-capacity edge server (ES) to improve efficiency. Most existing research focuses on inseparable task offloading to minimize response times or resource allocation to reduce energy consumption. However, task execution can be speeded up with excessive computing and network resources, it will increase energy consumption and incur unnecessarily high costs. Besides, complex applications like autonomous driving often partition sequential tasks to improve performance, necessitating a joint optimization of sequential task offloading and multi-resource allocation. In this paper, we introduce a Stackelberg game-based framework to model the interplay between these elements.EDs, acting as leaders, determine the offloading breakpoints of sequential tasks and the locality for processing. TheES, as the follower, uses the Karush-Kuhn-Tucker (KKT) conditions and a Boundary-constrained quasi-Particle Swarm Optimization (Bc-qPSO) algorithm to refine computing and network resource allocation, aiming to reduce system costs effectively. Our simulations show that the proposed algorithms reduce cost by approximately 10%-20% compared to traditional methods, highlighting their potential for improving the efficiency of edge computing systems.

As a promising distributed computing paradigm, federated learning (FL) has been widely applied to industrial manufacturing. To enhance the safety of FL system against growing privacy attacks, secret sharing techniques are introduced to realize secure model aggregation. However, the huge extra transmission in vanilla secret sharing aggravates the inherent communication bottleneck of FL system in industrial cellular networks. To reduce transmission overhead, this article proposes a lightweight FL system based on rotated partial secret sharing (FedRPS), which divides industrial terminals into several subgroups and allows them to alternately share partial local secrets. Our FedRPS ensures the global model to converge at a high rate with incomplete local parameters. Moreover, we theoretically analyze the convergence error of FedRPS, and discuss the impact of unbalanced data in subgroups on model convergence. To further improve global model performance under non-IID settings, we design a multiparty greedy grouping (MPGG) strategy to balance the training data in each subgroup. Finally, we conduct experiments on several datasets, which demonstrate our FedRPS trades for significant improvement on communication with only small loss of model accuracy. In addition, we also simulate the privacy attack scenarios to verify the better defense of FedRPS compared with previous complete secret sharing.

Telegram is a widely used instant messaging app that has gained popularity due to its high level of privacy protection. Telegram has standout social network features like channels, which are virtual rooms where only administrators can post and broadcast messages to all subscribers. However, these same features have also led to the emergence of problematic activities and a significant number of fake accounts. To address these issues, Telegram has introduced verified and scam marks for channels, but only a small number of official channels are currently marked as verified, and only a few fakes as scams. In this research, we conduct a large-scale analysis of Telegram by collecting data from 120,979 different public channels and over 247 million messages. We identify and analyze two types of channels: Clones and fakes. Clones are channels that publish identical content from another channel in order to gain subscribers and promote services. Fakes, on the other hand, are channels that impersonate celebrities or well-known services by posting their own messages. To automatically detect fake channels, we propose a machine learning model that achieves an F1-score of 85.45%. By applying this model to our dataset, we find the main targets of fakes are political figures, well-known people such as actors or singers, and services.

Many researchers primarily rely on modeling user interests for sequential recommendation. However, dynamic user behaviors often accompany unstable interaction histories, and modeling user interests alone is insufficient for comprehensive user features. Some studies notice the problem of interest drift, but they are usually limited to modeling at the item-level, unable to perceive the subtle changes at the feature-level. To this end, we propose a Stability-aware Preference Model (SAPM), which consists of three modules. The LSI module for extracting long and short-term interests, the FLC module for extracting feature-level candidate information, and the SAF module for fusing them according to the stability score. In particular, we propose a Multi-head GRU (MHGRU) structure in the LSI module, which is more efficient than the general GRU and has stronger expression ability. Through extensive experiments, our framework shows significant mitigation of the impact of unstable interactions. On the two real data sets, we improve MRR by 5.7% and 14.0% compared with the recent baselines. Moreover, we conduct an in-depth analysis of user interaction stability and obtain several interesting findings that can benefit future studies.

In federated learning, multiple rounds of communication are involved between clients and the server to train a global model. The extensive model updates transmitted during the training lead to significant communication costs. Previous methods usually employ quantization or sparsification to compress model updates. However, the lossy compression leads to a decline in accuracy, it is challenging to strike a balance between communication efficiency and model accuracy. Meanwhile, due to the data heterogeneity, local updates among different clients are biased towards each other. Employing the same compression ratios for each local updates will further degrade the model accuracy. To achieve the trade-off between communication efficiency and model accuracy, we propose FedDAC, a Dual Adaptive Compression method in heterogeneous federated learning. In the local computation phase, the loss queue is adopted to detect the convergence trends within each client. FedDAC can then dynamically quantify model updates and allow for various compression ratios among heterogeneous clients. In the global aggregation phase, FedDAC can determine the fluctuations in training based on the similarity between clients and the server, thereby adjusting the sparsity ratio flexibly. To alleviate the reduction in model accuracy caused by lossy compression, we introduce residual updates in the local computation and global aggregation phases to maintain model accuracy. Experiment results show that compared with one-way compression methods NAGC and AdaQuantFL, FedDAC can maintain comparable accuracy while the accumulated communication volume is reduced by about 29.6 times, and 22.8 times, respectively. Moreover, the global model accuracy of FedDAC surpasses the two-way compression method T-FedAvg by about 2.4%, and the accumulated communication volume is about 2.5 times lower than T-FedAvg.

In streaming federated learning, where data on each client is received in the form of a data stream, the distribution of data on the clients has a significant impact on the performance of the federated learning model. The continuous influx of streaming data on the clients leads to real-time changes in the local data distribution, which in turn affects the performance of the federated learning model. Furthermore, the heterogeneity in data distribution among clients exacerbates this impact. In this paper, to address these challenges, we propose a Data-aware Probabilistic Client Sampling scheme (DPCS) for selecting appropriate clients to participate in model training in each round of federated learning. DPCS begins with a method for real-time monitoring of local data distributions on the clients. Based on these observations, the central server adopts a probability-based client sampling strategy. Through extensive experimentation, we demonstrate that our client sampling scheme offers higher timeliness and enhances the performance of federated learning compared to traditional methods.

The rapid development of deep learning technologies and the widespread deployment of sensing devices have brought considerable attention to Internet of Things (IoT). The smart sensing application is one of the popular applications in IoT. Personalized Federated Learning (pFL) is a replacement to traditional Federated Learning (FL) to tackle the statistical heterogeneity of clients’ private datasets (e.g., non-Independent and Identically (non-IID) data). However, existing pFL methods encounter two challenges in smart sensing applications: a) the global model preference, causing poor global model performance for minority classes on sensing device data. b) the dynamic role differences in each hierarchy of the layer-stacked deep learning model that needs to be considered. Jointly considering these challenges, we present a novel edge-cloud enabled pFL framework named pFL-Sensing for smart sensing applications. Specifically, the sensing device serves as an edge server. Each edge server produces a customized model through two phases: a model training phase and a model aggregation phase. In the model training phase, we design a novel loss function to alleviate the issue of the global model preference. In the model aggregation phase, hierarchical aggregation and an Adaptive Weight Calculation (AWC) mechanism are proposed to capture the dynamic role differences of model hierarchies. We perform simulation experiments on real image and text classification tasks. Experimental results show that pFL-Sensing demonstrates higher classification accuracy than advanced pFL baselines.

Computing power network has emerged as an attractive technology to tackle the increasing demand for computational resources in cloud networking. In this paper, we study the problem of optimizing the formation of clusters of computing servers/nodes and also the task scheduling considering the games between the platform and computing nodes to maximize the platform profit. We formulate this problem as an integer programming problem. We propose a deep reinforcement-learning based server clustering and auction-based task scheduling algorithm working at different time scales to solve the problem. The deep reinforcement learning-based server clustering algorithm works at a large time scale to optimize the sizes and compositions of different clusters based on temporal and spatial distribution of the tasks and also their characteristics. The auction-based task scheduling algorithm works at a small time scale to match the tasks with the clusters while satisfying the QoS requirements of tasks so as to maximize the profit of the platform. Extensive simulations are conducted to evaluate the performance of proposed algorithm and the results show its high performance.

In real Internet of Things (IoT) environments, IoT devices vary widely in data types and needs. IoT devices participating in Personalized Federated Learning (PFL) all have their own unique data characteristics, but there exist some similarities. Previous work enhances Personalized Local Models (PLMs)’ performance by clustering IoT devices’ PLM while ignoring dimensionality and communication volume, resulting in lower Global Model (GM) accuracy and PLMs’ performance. To this end, we propose a Personalized Federated Learning method based on Model Head Clustering (FedMHC). Specifically, FedMHC groups IoT devices with similar data characteristics and distributes different GMs to different device groups. FedMHC allows each IoT device to obtain a GM that best fits its local data characteristics and guides PLM training. FedMHC only clusters model head parameters on the server side. Thus, the edge server only needs to transmit the head parameters and a single shared feature extractor parameters during communication with IoT devices. The improvement can effectively address the issues of dimensionality and high communication volume. Experiments on CIFAR-100, Tiny-ImageNet, and AG News datasets demonstrate that FedMHC enhances the model accuracy by 1.79% and 5.9% in pathological heterogeneous scenarios, and by 1.43%, 0.92%, and 0.94% in practical heterogeneous scenarios, compared to the topperforming methods among 9 baselines.

Most intrusion detection systems (IDS) are vulnerable to novel attacks and struggle to maintain a balance between high accuracy and a low false positive rate. Furthermore, the relevant features of Distributed Denial of Service (DDoS) attacks in conventional networks may not necessarily apply to the Software-defined network (SDN) environment. Additionally, weak feature selection algorithms can omit critical parameters and result in significant data loss. Although earlier works on network flow analysis using Long Short-Term Memory (LSTM) show excellent ability, they fall short in obtaining deep features from network flow, resulting in lower accuracy. The emergence of Attention Mechanism(AM) and deep reinforcement learning (DRL) present a promising solution for intrusion detection and enhancing security in SDN. AM has the capability to assign varying weights to different network traffic features, enabling IDS to extract and emphasize more crucial information. This paper introduces DeepIDPS, a novel DRL-based network intrusion detection system utilizing a CNN-LSTM approach and Attention Mechanism specifically designed for SDN environments. DeepIDPS demonstrates an exceptional ability for continuous auto-learning within the network context, effectively identifying diverse forms of network intrusions while significantly augmenting both prevention and detection capabilities.

The registration of LiDAR point clouds between Connected Vehicles (CV s) is the basis for vision tasks such as multi-vehicle cooperative perception and high-precision scene reconstruction. Due to the huge amount of LiDAR point cloud data and the complexity of computation, the registration has high transmission latency and execution time. Fast and accurate point cloud registration between CV s is urgently required by safe driving applications. To achieve faster registration, we propose partitioning overlapping point clouds for registration in advance. Meanwhile, for higher accuracy, we incorporate more geometric features during the registration. This paper demonstrates the Echo Registration System (ERS) developed to increase the speed of existing point cloud registration methods. ERS employs an initial positioning based on point cloud distributions, a density-consistent partition strategy, and a virtual geometric feature fusion. We deploy ERS to accelerate multiple state-of-the-art registration methods. The comprehensive results show that ERS improves the overall running time of current state-of-the-art baselines by 5.7 x with 42.1 % registration recall gains.

In recent years, many crowdsourcing platforms have emerged, using the resources of recruited workers to perform diverse outsourcing tasks, where the video analytics attracts much attention due to its practical implications. For maximum profits, platforms carefully choose the workers and determine the video analytics configurations to ensure accuracy; meanwhile, workers possess the flexibility to tailor the configurations for their indivi-dual gains, which makes it hard for platforms to optimize their profits considering the platform-worker conflicts. In this paper, we design an incentive mechanism for Multi-leader game-based video Analytics upon CROwdsourcing, named MACRO, to over-come the above situation. Under that mechanism, we first formu-late the utility optimization problems for platforms and workers, respectively. We then propose a dual ascent-based method to op-timally determine the video analytics configurations for a multi-platform game, ensuring Pareto efficiency. Moreover, in the context of a multi-leader game involving platform-worker conflicts, we design an incentive function with its incentive factor update strategy and propose an ADMM-based approach for maximizing incentives that motivate workers to contribute to the platforms' profits. Rigorous proofs demonstrate the linear convergence of the MACRO to the multi-leader Stackelberg equilibrium. Trace-driven experiments show that MACRO improves the Pareto efficiency by 26.3%, outperforming other approaches.

In this paper, we investigate the competitive content placement problem in Mobile Edge Caching (MEC) systems, where Edge Data Providers (EDPs) cache appropriate contents and trade them with requesters at a suitable price. Most of the existing works ignore the complicated strategic and economic interplay between content caching, pricing, and content sharing. Therefore, we propose a joint Mean-Field Game framework for mobile edge Caching and Pricing (MFG-CP) in large-scale dynamic MEC systems, which can facilitate distributed optimal decision-making based on the mean-field game theory. Specifi-cally, we first formulate the competitive content placement issue among EDPs as a non-cooperative stochastic differential game. To significantly reduce the communication and computation complexity, we further devise a mean-field model to approximate the collective impact of all EDPs on caching, trading, and sharing, by which each EDP can quickly estimate some unknown information without considerable interactions. Then, we develop a distributed best response scheme based on iterative learning, enabling each EDP to solely customize its optimal caching strategy and pricing policy. Besides, we theoretically prove the existence of a unique MFG equilibrium. Finally, trace-driven simulations demonstrate the effectiveness of MFG-CP compared with some baselines.

Federated Learning (FL) is an emerging privacy-preserving distributed machine learning paradigm that enables numerous clients to collaboratively train a global model without transmitting private datasets to the FL server. Unlike most existing research, this paper introduces a Data-Driven FL system in Unmanned Aerial Vehicle (UAV) networks, named DDFL, which features an innovative three-layer architecture. In this architecture, UAVs, inherently lacking data, serve as mobile clients and are tasked with periodically collecting desired data from a group of Points of Interest (PoIs) for FL training. Meanwhile, a Base Station (BS) coordinates these UAVs to efficiently train a high-quality global model. Our objective is to determine a data collection strategy for each UAV to minimize the time required to meet the global model’s loss requirement within the constraint of energy consumption. Through theoretical analysis, we establish a bound for the convergence speed of DDFL and quantify the impact of collecting data from different PoIs on the global model’s loss function. Leveraging these analyses, we formulate the PoI selection problem as a novel two-stage Combinatorial Multi-Armed Bandit (CMAB) problem with multiple constraints. We then propose an Adaptive Two-stage CMAB-based algorithm, named FedATC, to jointly optimize the data collection route and UAV velocity. Extensive simulations demonstrate that FedATC significantly reduces the time required to achieve desired model quality compared to state-of-the-art algorithms.

In cross-silo federated learning (FL), organizations cooperatively train a global model with their local data. The organizations, however, may be heterogeneous in terms of data distributions. In such cases, FL might produce a biased global model that is not optimal for each organization. Then each organization faces several fundamental questions: should I join FL or just remain alone? If joining FL, which organizations should I cooperate with? In this work, we formulate a coalition formation game in cross-silo FL to help organizations choose proper cooperators. We first build an estimation method to predict personal model performance for each organization before FL starts, and we treat performance improvement as individual utility. With estimated utilities, we design a distributed coalition formation algorithm to find stable coalition structures and optimize social welfare at the same time. Our simulations based on MNIST and FMNIST datasets show that the estimation model can predict the sign of the utility correctly with a probability of 0.9 and has an average relative error of $30 \%$. With the above errors, the obtained coalition structure performs well from both perspectives of real social welfare and individual satisfaction.

Recent years have witnessed the unprecedented performance of convolutional networks in image super-resolution (SR). SR involves upscaling a single low-resolution image to meet application-specific image quality demands, making it vital for mobile devices. However, the excessive computational and memory requirements of SR tasks pose a challenge in mapping SR networks on a single resource-constrained mobile device, especially for an ultra-high target resolution. This work presents TileSR, a novel framework for efficient image SR through tile-granular parallel offloading upon multiple collaborative mobile devices. In particular, for an incoming image, TileSR first uniformly divides it into multiple tiles and selects the top-K tiles with the highest upscaling difficulty (quantified by mPV). Then, we propose a tile scheduling algorithm based on multi-agent multiarmed bandit, which attains the accurate offload reward through the exploration phase, derives the tile packing decision based on the reward estimates, and exploits this decision to schedule the selected tiles. We have implemented TileSR fully based on COTS hardware, and the experimental results demonstrate that TileSR reduces the response latency by 17.77-82.2% while improving the image quality by 2.38-10.57% compared to other alternatives.

In the context of mobile edge computing, achieving dynamic service caching and updating to guarantee the QoS of users and reduce system costs is a challenging problem. However, existing research still has certain deficiencies in considering the dynamic behavior of users and the limited storage resources of edge servers. To address this problem, this paper proposes three novel strategies for the different stages of service caching and updating to jointly optimize the delay and cost. At the initial service caching stage, we propose a caching strategy based on dynamic programming, taking into account the constraint of limited memory resources. Given the dynamic behavior of users, we formulate the joint optimization problem as a Markov Decision Process (MDP) and design a service extension strategy based on Q-learning at the service updating decision-making stage and a replacement strategy taking both the distribution of service replications and service access frequency into account at the service updating replacement stage to guarantee the QoS of users. We effectively tackle the challenges arising from the dynamic behavior of users and limited storage resources. Through extensive comparative experiments, our approach outperforms traditional strategies by significantly reducing user latency and system cost.

Local Area Networks (LANs), as interconnected networks, are susceptible to numerous security threats. Existing intrusion detection systems (IDS) heavily rely on large, fully-labeled datasets to have accurate detection, facing challenges when only a few malicious samples are available. In addition, previous studies have identified the deterioration of IDS’s performance when the test dataset deviates from the training dataset distribution. To mitigate these issues, we propose a Prototypical Network-based IDS within a meta-learning framework. Our method adopts a Few-Shot Learning (FSL) approach, aiming to distinguish and compare network traffic samples to classify them as either normal or malicious. Notably, our model not only identifies benign or malicious traffic but also accurately identifies the specific types of attacks. We evaluate the effectiveness of our approach in different scenarios for few-shot network intrusion detection using real-world network traffic data. Additionally, we conduct a comprehensive sensitivity analysis to assess the impact of key factors such as model hyperparameters, support set size, attack type distribution, and distance metrics within the prototypical network model.

The Internet of Things (IoT) is gathering paces in the new era of Industry 4.0, and the Digital Twin (DT) technology bridges the gap between the bursting amounts of data generated by IoT devices and the user requirements for real-time data processing. DT services maintain living digital models of physical objects, and a DT network enables comprehensive service provisioning with the global knowledge of a group of DTs. On the other hand, exposing serverless computing in network edges, the recent advances in Serverless Edge Computing (SEC) introduce new inspirations to the DT landscape that ensure fine-grained resource management and low network-wide delay of DT services. However, social relationships among IoT devices and DT data privacy impact the orchestration of DTs. In this paper, we design a differential privacy-based federated learning framework to build a DT network for DT services in response to user DT service requests in SEC, thereby enhancing the Quality of Services (QoS). To this end, we first formulate a novel social-aware problem for placing DTs in an SEC network, and show its NP-hardness. We then provide an Integer Linear Program (ILP) solution to the problem when the problem size is small; otherwise, we design an approximation algorithm with a provable approximation ratio. We finally evaluate the algorithm performance through simulations. Simulation results demonstrate the proposed algorithm is promising, which improves by no less than 21.1 % of the performance of benchmarks.

In this paper, we consider the large-scale Area Under the Precision-Recall Curve (AUPRC) maximization problem for the imbalanced data classification task. Existing optimization methods for AUPRC maximization only focus on the single-machine setting, which are not applicable to the distributed data. To address this problem, we propose a novel decentralized stochastic compositional gradient descent method for large-scale AUPRC maximization. Our theoretical analysis shows that it can achieve a better sample complexity 𝒪 (1/ϵ4) than 𝒪 (1/ϵ6) of existing decentralized methods, but has the same communication complexity 𝒪 (1/ϵ4). To further reduce the communication cost, we developed a novel communication-efficient decentralized stochastic compositional gradient descent method, whose communication complexity is improved to 𝒪 (1/ϵ4–4α) (where α ∈ (0,1/4)). To the best of our knowledge, this is the first work achieving such favorable sample and communication complexities. Finally, we conduct extensive experiments for imbalanced data classification and the empirical results confirm the superior performance of our proposed methods.

Many extensions of the Multi-Armed Bandit (MAB) problem were studied recently offering a strong theoretical basis for applications that require active learning. The Budget-Constrained and Deadline-Driven MAB with Delay (BD-MAB) variation introduces a novel scenario in which a player must pull from K arms, each associated with reward, delay, and cost distributions. To the best of our knowledge, this is the first work that combines budget constraints with time delays in an MAB problem. The model involves three phases. The first phase comprises pulling arms, while incurring random cost, and observing rewards. This phase ends when the budget is depleted. The second phase includes observing some of the delayed rewards. The third phase starts at a fixed termination deadline, marking the end of reward observation, where any rewards returned after this phase are not observed and are considered dead. We present a novel solution to this problem by developing a new Upper Confidence Bound (UCB)-based algorithm. The name of the algorithm is Budget-Constrained and Deadline-Driven UCB with Delay (BD-UCB) algorithm. We provide extensive regret analysis that confirms the efficiency of our approach in managing these complexities. Lastly, we provide numerical simulations which further demonstrate the effectiveness of our proposed solution.

Payment channel networks (PCNs) have emerged as a viable solution to the scalability problem of blockchain systems. In PCNs, two peers can open a payment channel to transfer funds without publishing every transaction to a global blockchain. Payments can be routed between two unchanneled peers via a payment path, i.e., a sequence of adjacent payment channels. The routing protocol is the core of a PCN, since it regulates path discovery for transaction senders and receivers. Lots of routing algorithms have been proposed, most of which, however, lack either utility, i.e., a low payment success rate, or privacy, i.e., leaking channel balances. In this work, we first identify the utility-privacy tradeoff in existing routing algorithms. Aiming to trade privacy for utility, we propose a noised PCN by introducing a probabilistic model on the publicly-revealed channel balances. This enables us to express the reliability of a successful transaction for a given channel/path, Channel owners are allowed to select their own noise mechanisms, which directly link to routing fees they will charge. We further apply utility theory to design routing algorithms which achieve a balance between utility and privacy. Extensive simulations using a Lightning Network (LN) simulator called CLoTH have been conducted to validate the improvement in the payment success rate of the proposed architecture under different settings.

Mobile crowdsensing (MC), an excellent solution to large-scale spatiotemporal data sensing problems, has recently received lots of attention from both industry and academia. In the MC system, any requester can acquire the sensing data for his points of interest (PoIs) by offering some payments to attract a group of mobile users capable of completing these PoI-related sensing tasks. However, the current MC work neglected three vital factors, more or less. First, they assume that these distributed users are mutually independent in MC, ignoring the social effects. Actually, the sensing data collected by one user may be corroborated by others’ sensing data, so-called information corroboration. Second, all rational and selfish users are inclined to gather to perform these tasks due to information corroboration. Meanwhile, they may be strategic about their participation levels to maximize profits. However, more similar sensing data will undoubtedly lower the information value, so any user has a tradeoff between gather and scatter. Third, although mobile users can obtain some payments, privacy issues may still prevent them from participating in MC. In this article, we propose a secure blockchain-assisted socially-aware MC framework by adopting the smart contract technique of Ethereum. For this framework, we further devise a two-stage Stackelberg game model to assist the requester (i.e., the leader in the game) in properly pricing each PoI-related sensing task, so that mobile users (i.e., the followers in the game) can exactly select their tasks and determine their participation levels. To analyze the game equilibrium, we extend the traditional Hessian matrix method to a multidimension case involving the multiuser multitask hyperspace setting. We conduct extensive experiments to prove the equilibrium and effectiveness of the proposed solution. We also implement a prototype and deploy the smart contract to an official Ethereum test network to demonstrate the practicability of the proposed framework.

The Internet of Medical Things (IoMT) stimulates the development of intelligent medical applications. As mental disorders become a global problem, emotion recognition has received widespread attention, as it can contribute to more comprehensive mental health monitoring and psychological assessment. Physiological signal-based emotion-aware monitoring is a particularly promising application due to its noninvasive and objective data collection. Recently, multimodal emotion recognition has been enhanced with wireless body area network (WBAN) access to IoMT, where wireless medical sensors are interconnected and abundant signals are acquired conveniently. However, how to synthesize these multisource physiological signals to facilitate emotion recognition is a challenging problem due to their heterogeneity and interference. To solve this problem, we propose a real-time differential multimodal transformer (Diff-MT), where the main components are the differential hyperinformation extraction (DHE) module, the multimodal global cross-attention encoder (MGCE), and the difference-augmented feature fusion (DFF). Ultimately, we endow the system with emotional awareness and distribute the state to IoMT devices. Extensive experiments demonstrate that the proposed Diff-MT exhibits superior performance compared to existing methods on the WESAD and DEAP datasets and is appropriate for IoMT-based healthcare.

Mobile Crowd Sensing (MCS), as a promising sensing paradigm, significantly relies on wireless communication networks and widely distributed mobile workers to capture data from the surroundings. However, the positioning-dependent nature of most MCS tasks often requires workers to embed their positionings in reports, which may result in privacy leakage and a decline in their participation enthusiasm. Considering workers’ diverse perceptions of positioning privacy, in this paper we propose the Lightweight Positioning-Privacy Protection Scheme with Double-Layer Incentives (L3P-DLI) to meet their personalized privacy requirements in an efficient and low-cost way while stimulating their participation. To the best of our knowledge, this scheme is the first attempt to employ proxy forwarding to protect workers’ sensitive positionings while ensuring high-quality sensing results. Moreover, our double-layer incentivizing mechanism is elaborately designed to motivate workers to actively participate or serve as proxies. Specifically, the bidirectional auction between data collectors and proxies can safeguard the security of data collectors, and compensate for the potential privacy leakage cost of proxies helping to forward data. Additionally, the reverse auction mechanism enables the platform to reward recruited workers to compensate for their various costs. Extensive experiments conducted on real-world datasets validate that L3P-DLI effectively preserves workers’ positioning privacy while maximizing their income to encourage participation.

Graph Convolutional Networks (GCNs) have been widely used in skeleton-based human action recognition and have achieved promising results. However, current GCN-based methods are limited by their inability to refine semantic-guided joint relations and perform adaptive multi-scale analysis. These limitations impair their performance, particularly for analogical actions involving the interaction of the same body parts (e.g., drinking water and eating) as well as deficient actions with limited spatial-temporal information (e.g., subtle action writing and transient action sneezing). To solve these problems, we propose Part-level Refined Spatial Graph Convolution (PR-SGC) and Scale-aware Temporal Graph Convolution (Sa-TGC) for optimal action representation. The PR-SGC divides the skeleton into body parts and embeds this high-level semantics to refine the physical adjacency matrix. The Sa-TGC leverages the dynamic scale-aware mechanism to extract context-dependent multi-scale features. On this basis, we develop a novel Scale-aware Graph Convolutional Network with Part-level Refinement (SaPR-GCN), which is on par with state-of-the-art benchmarks on NTU RGB+D 60, NTU RGB+D 120, and NW-UCLA datasets.

The huge values created by Big Data and the recent advances in cloud computing have been driving data from different sources into cloud repositories for comprehensive query services. However, cloud-based data fusion makes it challenging to verify if an untrusted server faithfully integrates data and executes queries or not. This is even harder for range-aggregate queries that apply aggregate operations on data within given ranges. In this paper, we propose a query authentication scheme, named${\sf MARS}$, enabling a user to efficiently authenticate range-aggregate queries on multi-source data. Specifically,${\sf MARS}$creates a VG-tree by subtly integrating Expressive Set Accumulator into a multi-dimensional G-tree while signing the root digest with a multi-source aggregate signature scheme. Compared with previous solutions,${\sf MARS}$has the following merits: (1)Practicality.Instead of treating range and aggregate queries separately, the user can directly verify the statistical result of selected data. (2)Scalability.Instead of authenticating the individual result from each source, the user can perform an aggregative validation on the integrated result from multiple sources. The experimental results demonstrate the effectiveness of MARS. For large-scale data fusion, the user-side verification time increases by only 103 ms as the amount of data sources increases by five times.

By decoupling the control plane and data plane in the software-defined network (SDN), the controller gains a comprehensive global view of the network. The SDN controller samples traffic from all switches to effectively manage data plane traffic. The sampling rate of flow traffic significantly impacts the accuracy of the controller’s decisions. While increasing the sampling rate is desirable for improved detection accuracy, it also escalates resource consumption on both switches and the controller. Hence, it is crucial to carefully manage sampling on switches to fine-tune anomaly detection accuracy. Existing flow sampling solutions often struggle to strike a balance between detection accuracy, sampling rate, and overhead. To address this challenge, we propose a robust cybersecurity framework for anomaly detection on SDNs through traffic flow inspection. Our proposed framework, Cyber-AnDe, integrates adaptive distributed sampling (ADS) with a Reinforcement Learning (RL) agent to enhance anomaly detection accuracy while minimizing the increase in controller overhead. In our framework, the controller leverages information gathered from each sampled traffic flow to determine whether the flow’s state is malicious, suspicious, or benign based on underlying anomaly detection algorithms. Once the flow state is determined, the controller takes the appropriate action with the help of the RL agent. Through extensive simulations and SDN test-bed experiments, we confirm a significant improvement of up to 93% in network traffic-based anomaly detection compared to existing solutions.

With the wide adoption of blockchains in data-intensive applications, enabling verifiable queries over a blockchain database is urgently required. Aiming at reducing costs, previous solutions embed a small-sized authenticated data structure (ADS) in each block header, so that a user can verify search results without maintaining a full copy of blockchain databases. However, existing studies focus on exact queries with difficulty to guarantee the freshness of search results. In this article, we propose two frameworks, called$\mathsf{veffChain}$and$\mathsf{veffChain++}$, to realize freshness authentication of rich queries over blockchain databases. Specifically,$\mathsf{veffChain}$concerns about verifiable latest-$K$exact queries and employs RSA accumulator to generate constant-size ADSs;$\mathsf{veffChain++}$integrates RSA accumulator into the Trie tree to further authenticate latest-$K$fuzzy queries. For improved scalability, an adaptive keyword splitting (AKS) solution is proposed to enable ADSs to be incrementally updated. Compared with the state-of-the-art work, our frameworks have the following merits: (1)Freshness Guarantee. The user can efficiently retrieve the freshest data from a blockchain database in a verifiable way. (2)Flexibility. The user can specify different query patterns on demand to retrieve data as accurately as possible. The detailed security analysis and extensive experiments validate the practicality of our frameworks.

Due to the large-scale data streams produced by distributed terminals, hybrid-storage blockchain (HSB) that combines on-chain and off-chain storages has emerged as a promising solution for secure data storage in decentralized applications. Because all the raw data is outsourced to an untrusted service provider (SP), existing solutions suggest to utilize an on-chain authenticated data structure (ADS) to verify query results retrieved off-chain. However, existing solutions support onlycoarse-grained authenticationmaking a user abandon all the query results once the validation fails. In this paper, we focus on realizingfine-grained authenticationfor range queries, enabling a user to distinguish authentic data from falsified results. Considering the heavy gas consumption of on-chain storage, we propose two multi-dimensional parity-based verification (MPV) schemes with a trade-off between off-chain and on-chain efficiencies. Our main idea is to design an accumulator-based ADS to summarize well-designed verifiable hypercubes, so that fake results can be quickly located by combining multi-dimensional faces failed validation. Compared with previous solutions, our MPV schemes allow a user to make efficient use of query results by filtering out errors, and thus have higher data utility. The detailed security analysis and extensive experiments demonstrate the security and effectiveness of our MPV schemes, respectively.

Deep Neural Networks (DNNs) have been widely deployed in mobile applications. DNN inference latency is a critical metric to measure the service quality of those applications. Collaborative inference is a promising approach for latency optimization, where partial inference workloads are offloaded from mobile devices to cloud servers. Model partition problems for collaborative inference have been well studied. However, little attention has been paid to optimizing offloading pipeline for multiple DNN inference jobs. In practice, mobile devices usually need to process multiple DNN inference jobs simultaneously. We propose to jointly optimize the DNN partitioning and pipeline scheduling for multiple inference jobs. We theoretically analyze the optimal scheduling conditions for homogeneous chain-structure DNNs. Based on the analysis, we proposed near-optimal partitioning and scheduling methods for chain-structure DNNs. We also extend those methods for general-structure DNNs. In addition, we extend our problem scenario to handle heterogeneous DNN inference jobs. A layer-level scheduling algorithm is proposed. Theoretical analyses show that our proposed method is optimal when computation graphs are tree-structure. Our joint optimization methods are evaluated in a real-world testbed. Experiment results show that our methods can significantly reduce the overall inference latency of multiple inference jobs compared to partition-only or schedule-only approaches.

Mobile Edge Computing (MEC) is a promising solution to enhance the computing capability of resource-limited networks. A fundamental problem in MEC is efficiently offloading tasks from user devices to edge servers. However, there still exists a gap to deploy in real-world environments: 1) traditional centralized approaches needs complete information of edge network, ignoring the communication costs generated by synchronization, 2) previous works do not consider concurrent computation on edge servers, which may cause dynamic changes in the environment, and 3) the scheduling algorithm should deliver individualized decisions for different users independently and with high efficiency To solve this mismatch, we studied a multi-user task offloading problem where user devices make offloading decisions independently. We consider the concurrent execution of tasks and formulate a non-divisible and delay-aware task offloading problem to jointly minimize the dropped task ratio and long-term latency. We propose a decentralized task scheduling algorithm based on DRL that makes offloading decisions without knowing the information of other user devices. We employ Double-DQN, Dueling-DQN, Prioritized Replay Memory, and Recurrent Neural Network (RNN) techniques to improve the algorithm's performance. The results of simulation experiments show that our method can significantly reduce the long-term latency and dropped task ratio compared to the baseline algorithms.

Routing plays an essential role in ensuring normal and lasting operation of wireless body area networks (WBANs). However, existing routing schemes cause inefficient and unbalanced energy dissipation, which contributes to premature death of some nodes and high temperature within a small area of the body. In this article, we propose an adaptive time-varying routing (ATVR) protocol to address these issues. Unlike in conventional routing solutions, in our protocol a node may act as different roles (source node or relay node) and select different paths in disparate periods. This dynamic routing pattern helps to achieve a globally optimal routing solution. In ATVR, a node evaluation function and a path evaluation function are designed to reflect the node state and the path state respectively. Then, the path selection problem is transformed into a Hitchcock transportation problem, in which the nodes with worse node state act as source nodes (i.e., producers) and the nodes with better node state act as relay nodes (i.e., consumers). Then, this Hitchcock transportation problem is addressed by the AlphaBeta algorithm, in which the paths with less energy consumption and less path loss are selected to forward data. The experimental results show that our protocol has better performance in terms of energy consumption, network lifetime, and node temperature.

With the increasing awareness of secure data outsourcing, dynamic searchable symmetric encryption (DSSE) that enables searches and updates over encrypted data has begun to receive growing attention. Despite promising, existing DSSE schemes with forward and backward privacy are still hard to achieve authorized keyword searches on mobile devices while supporting secure and flexible updates. In this article, we propose a DSSE scheme, named$\mathsf{FLY_{++}}$based on a flexible index structure$\mathsf{Hybrid}$that incorporates the merits of inverted indexes and forward indexes while compacting the index size. Specifically,$\mathsf{FLY_{++}}$encrypts the newly added data with a fresh key and disperses previous keys into$\mathsf{Hybrid}$for forward privacy, while applying symmetric puncturable encryption (SPE) and a dual-key mechanism to realize backward privacy further. Compared with the state-of-the-art work,$\mathsf{FLY_{++}}$has the following advantages: (1)Authorized search. It dispenses with caching or re-encrypting search results, enabling a mobile device to search only designated keywords over the data outsourced before authorization. (2)Flexibility.It not only allows for sublinear search time, but also simultaneously supports fine-grained and coarse-grained updates of outsourced data. The detailed security analysis and extensive experiments conducted on a real dataset demonstrate the security and practicality of$\mathsf{FLY_{++}}$, respectively.

Parking violation is a common urban problem in major cities all over the world. Traditional approaches for detecting parking violations mainly rely on fixed deployed sensors and enforcement agencies, which suffer from high deployment costs and limited coverage. With the rapid development of mobile networks, Mobile CrowdSensing (MCS) has been an effective sensing paradigm. The crowdsensing data can help predict the future parking violation distribution, and the prediction results can provide guidance for user scheduling, i.e., sending the mobile users to patrol areas where many parking violation events may occur. Inspired by this idea, we propose a comprehensive data-driven crowdsensing framework, which incorporates the nested design of a generative model for spatial-temporal data and a user scheduling model. The generative model extracts parking violation hotspots via a data completion module and violation prediction module. Since crowdsensing data is usually temporally sparse and unevenly distributed, a data completion module is proposed to infer the missing statistics in unsensed areas. The violation prediction module then predicts the parking violation distribution. Given the predicted results, the deep reinforcement learning-based user scheduling model coordinates users to visit hotspots for violation detection. Iteratively, the newly collected data can be used to predict the future violation distribution. Finally, we conduct extensive simulations based on two real-world datasets from two large urban cities. The simulation verifies the prediction accuracy and scheduling effectiveness of the proposed framework compared with the baselines.

We consider a potential communication problem in national security, where wireless spy sensors with eavesdropping capability are strategically deployed around an area of interest. For counterintelligence, achieving secure communication by penetrating such a spy barrier is of great importance. In this paper, we first formulate the problem of barrier penetration routing against spy barriers consisting of strategically deployed wireless sensors. We point out that existing multi-path avoidance routing protocols cannot efficiently counteract collusion attacks, where connected adversaries collaborate with each other to compromise data packets. We propose a barrier penetration routing (BPR) protocol to securely penetrate the barrier of adversaries. In the protocol, a set of physically distanced paths are identified based on distance vectors as well as network-wide flooding. Then, each data packet encoded by XOR coding is routed via a different path. Unlike existing avoidance routing, the proposed scheme does not rely on the assumption that the adversary's locations are known. The simulation results demonstrate that the proposed BPR outperforms the baseline protocol as well as existing routing protocols in terms of secure delivery rate.

Deep convolutional neural network (NN)-based object detectors are not appropriate for straightforward inference on high-resolution videos at edge devices, as maintaining high accuracy often brings about prohibitively long latency. Although existing solutions have attempted to reduce on-device inference latency by selecting a cheaper configuration (e.g., choosing a more lightweight NN or scaling a frame to a smaller size before inference) or eliminating a background containing no object, they often ignore various high-resolution features and fail to optimize for those videos. We thus present AdaPyramid, a framework to reduce as much on-device inference latency as possible, especially for high-resolution videos, while achieving the accuracy demand approximately. We observe that the cheapest configuration to achieve the accuracy demand varies significantly across both different frames and different regions in a frame. The underlying reason is that object features (e.g., the location, size and category of objects) are more uneven in high-resolution videos, both temporally and spatially. Moreover, we observe that the object size presents a prominent hierarchical distribution in high-resolution frames. AdaPyramid thus partitions each frame hierarchically just like a pyramid and chooses a content-aware configuration for each region, which is adapted online based on the feedback. We evaluate the performance of AdaPyramid on a public dataset and our collected real-world videos. The obtained results show that under comparable accuracy to the state-of-the-art solutions, AdaPyramid can decrease inference latency by 40% on average, with up to 2.5× speed-up.

The advancements in information and communication technology have led to the emergence of innovative edge computing models that incorporate the computing power of vehicles into the energy sector. Electric vehicles (EVs), functioning as edge computing nodes, offer flexible computing offloading services for charging stations (CS). However, coordinating EV computing and charging should consider the interdependence with CS's specific computing requirements due to information asymmetry. Additionally, it is crucial to consider EV's charging demands and their social distance to computing tasks. In this context, it is natural to view EVs and CSs as self-interested prosumers who prioritize their individual utilities. To address the integration of strategic EV-CS interactions and uncertainties into the joint computing scheduling and energy trading, this paper proposes a parameterized deep Q-network-based reinforcement contract design framework, which employs a hybrid action space to design contracts that facilitate CSs in pairing computing tasks and charging resources with EVs. The objective is to incentivize EV participation and maximize long-term social welfare by incorporating incentive compatibility, individual rationality constraints, and capacity constraints into the contract design. Experimental results demonstrate that the proposed framework surpasses parameterized deep deterministic policy gradient-based and greedy-based contract designs, and achieves near-optimal solutions by solving deterministic optimizations.

With the wide spread of data-driven deep learning applications, a growing number of users outsource compute-intensive inference processes to the cloud. To protect inference privacy, Liu (INFOCOM 2022) proposed two steganography-based solutions, named GHOST and GHOST+, relying on the mobile-cloud collaborative framework, where the mobile device hides sensitive images into public cover images before feature extraction, while launching adversarial attacks on the cloud-side deep neural network (DNN) to obtain desired results. Although both solutions demonstrate significant advantages in private deep learning, they suffer from limited practicality; since the inference accuracy decreases sharply as the hiding ratio increases. To address this, we propose two improved solutions, IGHO and IGHO+, which ensure high inference accuracy even when abundant sensitive images need to be hidden. Specifically, IGHO as the improved version of GHOST proposes two feature fusion methods, feature synthesis and pixel synthesis, to preprocess cover images, making the poisoned DNN learn hidden sensitive features better, while IGHO+as the improved version of GHOST+designs a novel feature mining generative adversarial network (FMGAN) to craft adversarial perturbations highly robust against variable sensitive types. Experimental results show that the proposed solutions highly improve the practicality of GHOST and GHOST+.

Mobile CrowdSensing (MCS) is a promising sensing paradigm that leverages users’ mobile devices to collect and share data for various applications. A key challenge in MCS is task allocation, which aims to assign sensing tasks to suitable users efficiently and effectively. Existing task allocation approaches are mostly centralized, requiring users to disclose their private information and facing high computational complexity. Moreover, centralized approaches may not satisfy users’ preferences or incentives. To address these issues, we propose a novel distributed task allocation scheme based on route navigation systems. We consider two scenarios: time-tolerant tasks and time-sensitive tasks, and formulate them as potential games. We design distributed algorithms to achieve Nash equilibria while considering users’ individual preferences and the platform’s task allocation objectives. We also analyze the convergence and performance of our algorithm theoretically. In the time-sensitive task scenario, the problem becomes even more intricate due to temporal conflicts among tasks. We prove the task selection problem is NP-hard and propose a distributed task selection algorithm. In contrast to existing distributed approaches that require users to deviate from their regular routes, our method ensures task completion while minimizing disruption to users. Trace-based simulation results validate that the proposed algorithm attains a Nash equilibrium and offers a total user profit performance closely aligned with that of the optimal solution.

Mobile CrowdSensing (MCS) is a widely adopted approach that involves engaging mobile users to collaboratively perform diverse sensing tasks. In Sparse CrowdSensing, the completion of data from partially-sensed sources plays a pivotal role in urban sensing applications. This process is essential as it enables efficient data representation, enhances urban analysis capabilities, and ultimately facilitates informed city planning decisions. By leveraging the power of mobile users, Sparse CrowdSensing contributes to the comprehensive understanding of urban environments, enabling effective utilization of data for optimizing urban infrastructure and fostering sustainable urban development. To achieve accurate completion results, previous methods usually utilize the universal similarity and conventional tendency while incorporating only a single dataset to infer the full map. However, in real-world scenarios, there may exist many kinds of data (inter-data), that could help to complement each other. Moreover, for each kind of data (intra-data), there usually exists a few but important spatiotemporal fracture data which caused by the special events (e.g. data loss, equipment failure, etc.), which may behave in a different way as the statistical patterns. Thus, it is an essential task to consider spatiotemporal fracture data caused by the special cases in spatiotemporal data inference, especially using both intra- and inter-data, because of the following challenges: 1) the sparsity of the sensed data, 2) the complex spatiotemporal relations and 3) the uncertain scale of a spatiotemporal fracture. To this end, focusing on the large-scale spatiotemporal fracture, we propose a data completion method that exploits both intra- and inter-data correlations for enhancing performance. Specifically, for the purpose of generating spatiotemporal fracture data, there isStackedGenerativeMatrixCompletion(SGMC)by combining previous Stacked Deep Matrix Factorization (SDMF) and Generative Adversarial Networks (GANs), which improves a lot. Along this line, we extract the features of spatiotemporal data and further efficiently complete and predict the unsensed data. Finally, we conduct both qualitative and quantitative experiments on two different datasets, and the results demonstrate that the performance of our method outperforms the state-of-the-art baselines.

In this paper, we focus on the privacy preserving mechanism design for crowdsourced Federated Learning (FL), where a requester can outsource its model training task to some workers via an FL platform. A potential way to preserve the privacy of workers' local data is to leverage Differential Privacy (DP) mechanisms on local models. However, most of these studies cannot allow workers to dominate their own privacy protection levels by themselves. Thus, we propose a Personalized Privacy Preserving Mechanism, called P3M, to satisfy the heterogeneous privacy needs of workers, which consists of two parts. The first part includes a personalized privacy budget determination problem. We model it as a two-stage Stackelberg game, derive the personalized privacy budget for each worker and the optimal payment for the requester, and prove that they form a unique Stackelberg equilibrium. Second, we design a dynamic perturbation scheme to perturb model parameters. Through the theoretical analysis, we prove that P3M satisfies the desired DP property, and derive the bounds of the variance of average perturbed parameters and the convergence upper bound. This demonstrates that the global model accuracy can be controllable and P3M is endowed with the satisfactory convergence performance. In addition, we extend our problem to the scenario where the total privacy budget of all workers is limited, so as to prevent some workers from setting exorbitant privacy budgets. Under the privacy constraint, we re-determine the personalized privacy budget for each worker. Finally, exhaustive simulations of P3M are conducted based on real-world datasets, and the experimental results corroborate its effectiveness and practicability.

With the explosive spread of smart mobile devices, Mobile CrowdSensing (MCS) has been becoming a promising paradigm, by which a platform can coordinate a group of workers to complete large-scale data collection tasks using their mobile devices. In this paper, we investigate the incentive mechanism design in MCS systems, taking the freshness of collected data and social benefits into consideration. First, the Age of Information (AoI) metric is introduced to measure the freshness of data. Then, we model the incentive mechanism design with AoI guarantees as an incomplete information two-stage Stackelberg game with multiple constraints. Next, we consider the scenario that all participants share the public utility function parameters of the Stackelberg game. By deriving the optimal remuneration paid by the platform and the optimal data update frequency for each worker, and proving the existence of a unique Stackelberg equilibrium, we propose an AoI-guaranteed Incentive Mechanism (AIM) that enables the platform and all workers to maximize their utilities simultaneously. Furthermore, we extend AIM to a general scenario where each participant has no prior knowledge of the utility function parameters of the game. By resorting to the Deep Reinforcement Learning (DRL) technique and modeling the two-stage Stackelberg game as a Markov decision process, we propose a DRL-based Incentive Mechanism (DIM) with AoI guarantees, which makes each participant effectively seek its optimal strategy through trial and error. Meanwhile, the system can guarantee that the AoI values of all data uploaded to the platform are not larger than a given threshold. Finally, numerical experiments on real-world traces are conducted to validate the efficacy and efficiency of AIM and DIM.

The usage of live streaming services has led to a substantial increase in live video traffic. However, the perceived quality of experience of users is frequently limited by variations in the upstream bandwidth of streamers. To address this issue, several adaptive bitrate (ABR) algorithms have been developed to mitigate bandwidth variations. Nevertheless, the ability of users to enjoy high-quality live streams remains limited. While neural-enhanced approaches, such as super-resolution, offer significant quality improvements, frame-oriented super-resolution leads to excessive inference delay that violates the real-time feature of live streaming. In response, we propose ViChaser, which examines block-oriented super-resolution for live streaming. ViChaser performs neural super-resolution on potential blocks of interest in the media server, corresponding to the user’s viewpoint, and uses online learning to adapt to the dynamic content of the video. Additionally, ViChaser utilizes the Lyapunov framework to efficiently allocate uplink bandwidth for original low-quality live video and high-quality labels. The experimental results demonstrate that ViChaser achieves 1.2–1.5 dB higher video quality in Peak-Signal-to-Noise-Ratio than WebRTC and increases processing speed by 11–16 fps relative to LiveNAS.

The Mobile Edge Computing (MEC) paradigm gives impetus to the vigorous advancement of the Internet of Things (IoT), through provisioning low-latency computing services at network edges. The emerging digital twin technique has been explosively growing in the IoT community, which bridges the gap between physical objects and their digital representations in an MEC network, enabling real-time monitoring and analysis, simulations on the dynamics of systems, accurate predictions on behaviours of objects, and optimization on network resource allocation. In this paper, we consider AoI-aware query services in an MEC network empowered by digital twin technology for diverse IoT applications. We aim to maximize the weighted sum of the accumulative freshness of query results measured by the Age of Information (AoI) and the total query service delay of admitted requests. To this end, we first formulate a novel minimization problem that explores nontrivial trade-offs between the two conflicting optimization objectives: the freshness of query results and service delays, and we show the NP-hardness of the problem. Then, we propose an approximation algorithm with a provable approximation ratio for the problem, at the expense of bounded computing capacity violations. We also develop a heuristic for the problem without any capacity violations. We finally evaluate the performance of the proposed algorithms via simulations. The simulation results demonstrate that the proposed algorithms are promising, and outperform the comparison benchmarks.

Federated Learning (FL) is a distributed machine learning framework in parallel and distributed systems. However, the systems’ Non-Independent and Identically Distributed (Non-IID) data negatively affect the communication efficiency, since clients with different datasets may cause significant gaps to the local gradients in each communication round. In this article, we propose a Federated Vectorized Averaging (FedVeca) method to optimize the FL communication system on Non-IID data. Specifically, we set a novel objective for the global model which is related to the local gradients. The local gradient is defined as a bi-directional vector with step size and direction, where the step size is the number of local updates and the direction is divided into positive and negative according to our definition. In FedVeca, the direction is influenced by the step size, thus we average the bi-directional vectors to reduce the effect of different step sizes. Then, we theoretically analyze the relationship between the step sizes and the global objective, and obtain upper bounds on the step sizes per communication round. Based on the upper bounds, we design an algorithm for the server and the client to adaptively adjusts the step sizes that make the objective close to the optimum. Finally, we conduct experiments on different datasets, models and scenarios by building a prototype system, and the experimental results demonstrate the effectiveness and efficiency of the FedVeca method.

In this paper, we investigate the privacy-preserving task push problem with unknown popularity in Spatial Crowdsourcing (SC), where the platform needs to select some tasks with unknown popularity and push them to workers. Meanwhile, the preferences of workers and the popularity values of tasks might involve some sensitive information, which should be protected from disclosure. To address these concerns, we propose a Privacy Preserving Auction-based Bandit scheme, termed PPAB. Specifically, on the basis of the Combinatorial Multi-armed Bandit (CMAB) game, we first construct a Differentially Private Auction-based CMAB (DPA-CMAB) model. Under the DPA-CMAB model, we design a privacy-preserving arm-pulling policy based on Diffie-Hellman (DH), Differential Privacy (DP), and upper confidence bound, which includes the DH-based encryption mechanism and the hybrid DP-based protection mechanism. The policy not only can learn the popularity of tasks and make online task push decisions, but also can protect the popularity as well as workers’ preferences from being revealed. Meanwhile, we design an auction-based incentive mechanism to determine the payment for each selected task. Furthermore, we conduct an in-depth analysis of the security and online performance of PPAB, and prove that PPAB satisfies some desired properties (i.e., truthfulness, individual rationality, and computational efficiency). Finally, the significant performance of PPAB is confirmed through extensive simulations on the real-world dataset.

Opportunistic network-based mobile crowdsourcing (MCS) outsources location-based human tasks to a crowd of workers, where workers with mobile devices opportunistically have contact with the server. While a number of task assignment algorithms have been proposed for different objectives, real-timeness is not considered. In this article, we are interested in real-time MCS (RT-MCS), in which tasks can be generated at any time step, and task assignment is performed in real-time. We first model an abstract RT-MCS and then instantiate the real-time task assignment problem for opportunistic network-based RT-MCS. A generic real-time task assignment (RTA) algorithm is designed based on the principle of the greedy approach, where each task is assigned to the best worker with the highest expected completion probability. To understand the fundamental performance issues, we formulate closed-form solutions for task completion probability as well as delay. In addition, we identify the critical condition that illuminates the busy state and the not-busy state of an RT-MCS. Furthermore, the analytical and simulation results demonstrate that our analysis yields close approximation of simulation results.

The vigorous development of IoT technology has spawned a series of applications that are delay-sensitive or resource-intensive. Mobile edge computing is an emerging paradigm that provides services between end devices and traditional cloud data centers to users. However, with the continuously increasing investment of demands, it is nontrivial to maintain a higher quality-of-service (QoS) under the erratic activities of mobile users. In this paper, we investigate the service provisioning and updating problem under the multiple-users scenario by improving the performance of services with long-term cost constraints. We first decouple the original long-term optimization problem into a per-slot deterministic one by using Lyapunov optimization. Then, we propose two service updating decision strategies by considering the trajectory prediction conditions of users. Based on that, we design an online strategy by utilizing the committed horizon control method looking forward to multiple slots predictions. We prove the performance bound of our online strategy theoretically in terms of the trade-off between delay and cost. Extensive experiments demonstrate the superior performance of the proposed algorithm.

Recently, in response to the low efficiency and high transmission latency of traditional centralized content delivery networks, especially in congested scenarios, edge caching has emerged as a promising method to bring content caching closer to the edge of the network. However, traditional content delivery methods might still lead to low utilization of cache resources. To tackle this challenge, this paper investigates a content recommendation-based edge caching method in multi-tier edge-cloud networks while considering content delivery and cache replacement decisions as well as bandwidth allocation strategies. First, we consider a multi-tier edge caching-enabled content delivery network architecture combined with a content recommendation system and formulate the optimization problem with the objective of minimizing long-term content delivery delay and maximizing cache hit rate. Second, considering time-varying system environments and uncertain content demands, we approximate the optimization process of content delivery and cache replacement for each agent as a Partially Observable Markov Decision Process (POMDP) and propose a single-agent Deep Deterministic Policy Gradient (DDPG)-based method. Subsequently, we extend the POMDP to a multi-agent scenario. To address the issue of agents converging to local optima and establish more personalized models, we propose a Federated Distributed DDPG-based method (FD3PG) to solve the corresponding problem in a multi-agent system. Finally, simulation results demonstrate that the proposed FD3PG achieves lower delivery delay and higher cache hit rate compared with other baselines in various scenarios. Specifically, compared with FADE, MADRL, and DDPG, FD3PG achieves a significant decrease in average delivery delay, approximately 10%, 11%, and 35% on the Synthetic dataset, and 12%, 14%, and 48% on the MovieLens Latest Small dataset, respectively.

In the context of mobile edge computing, efficiently deploying microservices to reduce finish time and enhance user service quality is a challenging task. However, existing research still has certain deficiencies in considering microservice deployment within edge server clusters and communication link constraints. To address this issue, we propose three microservice deployment strategies by offering flexibility and adaptability for various application scenarios. We initially consider two straightforward scenarios: one with unlimited storage resources under the bandwidth constraint, and the other with unlimited bandwidth resources under the storage constraint. For each of these two scenarios, we introduce a novel enhanced graph construction method and design two optimal solutions. For Scenario 3, which involves complex constraints on server capacity, computational capability, and communication resources, we present an optimization method based on main path partitioning and the simulated annealing algorithm. We effectively tackle challenges arising from server capacity, computational capability, and communication resource limitations. Across multiple experimental setups, our approach significantly improves microservice deployment efficiency and overall performance compared to traditional strategies.

Deep Neural Network (DNN) models have been widely utilized in various applications. However, the growing complexity of DNNs has led to increased challenges and prolonged training durations. Despite the availability of high-performance computing systems, certain DNNs still require several days for successful training. This study aims to address this issue by proposing a method for significantly reducing the training time of deep learning models while maintaining test accuracy. Existing approaches primarily concentrate on optimizing training efficiency through computational and communication overlap/scheduling. In contrast, this research takes a step further by inspiring transfer learning. Transfer learning is a useful way to quickly retrain a model on new data without having to retrain the entire network. During transfer learning, the first layers of the network are frozen while leaving the end layers open to modification. By doing so, computation and communication requirements in these frozen layers are eliminated. This intelligent approach involves freezing some of the specific DNN layers and allocating resources to the remaining active layers during the training process, thereby minimizing DNN training time. To achieve this objective, we propose an intelligently freezing DNN using pipeline parallelism. Through trace-based simulation results, our scheme has demonstrated its effectiveness in efficiently reducing the time cost of a training iteration.

Telegram is a widely used instant messaging app that has gained popularity due to its high level of privacy protection and social network features like channels, which are virtual rooms where only administrators can post and broadcast messages to all subscribers. However, these same features have also led to the emergence of problematic activities and a significant number of fake accounts. To address these issues, Telegram has introduced verified and scam marks for channels, but only a small number of official channels are currently marked as verified, and only a few fakes as scams.In this research, we conduct a large-scale analysis of Telegram by collecting data from 120,979 different public channels and over 247 million messages. We identify and analyze fake channels on Telegram. To automatically detect fake channels, we propose a machine learning model that achieves an accuracy of 85.49%. By applying this model to our dataset, we find the main targets of fakes are political figures, well-known people such as actors or singers, and services.

Stochastic Multi-Armed Bandit (MAB) has recently been studied widely due to its vast range of applications. The classic model considers the reward of a pulled arm to be observed after a time delay that is sampled from a random distribution assigned for each arm. In this paper, we propose an extended framework in which pulling an arm gives both an instant (short-term) reward and a delayed (long-term) reward at the same time. The distributions of reward values for short-term and long-term rewards are related with a previously known relationship. The distribution of time delay for an arm is independent of the reward distributions of the arm. In our work, we devise three UCB-based algorithms, where two of them are near-optimal-regret algorithms for this new model, with the corresponding regret analysis for each one of them. Additionally, the random distributions for time delay values are allowed to yield infinite time, which corresponds to a case where the arm only gives a short-term reward. Finally, we evaluate our algorithms and compare this paradigm with previously known models on both a synthetic data set and a real data set that would reflect one of the potential applications of this model.

Mobile CrowdSensing (MCS) is a mobile computing paradigm, through which a platform can coordinate a crowd of workers to accomplish large-scale data collection tasks using their mobile devices. Information freshness has attracted much focus on MCS research worldwide. In this paper, we investigate the incentive mechanism design in MCS systems that take the freshness of collected data and social benefits into concerns. First, we introduce the Age of Information (AoI) metric to measure the freshness of data. Then, we model the incentive mechanism design with AoI guarantees as a novel incomplete information two-stage Stackelberg game with multiple constraints. Next, we derive the optimal strategies of this game so as to determine the optimal reward paid by the platform and the optimal data update frequency for each worker. Moreover, we prove that these optimal strategies form a unique Stackelberg equilibrium. Based on the optimal strategies, we propose an AoI-Aware Incentive (AIAI) mechanism for the MCS system, whereby the platform and all workers can maximize their utilities simultaneously. Meanwhile, the system can ensure that the AoI values of all data uploaded to the platform are not larger than a given threshold to achieve high data freshness. Extensive simulations on real-world traces are conducted to demonstrate the significant performance of AIAI.

Multi-Armed Bandit (MAB) that solves the sequential decision-making to the prior-unknown settings has been extensively studied and adopted in various applications such as online recommendation, transmission rate allocation, etc. Although some recent work has investigated the multi-agent MAB model, they supposed that agents share their bandit information based on social networks but neglected the incentives and arm-pulling budget for heterogeneous agents. In this paper, we propose a transaction-based multi-agent MAB framework, where agents can trade their bandit experience with each other to improve their total individual rewards. Agents not only face the dilemma between exploitation and exploration, but also decide to post a suitable price for their bandit experience. Meanwhile, as a buyer, the agent accepts another agent whose experience will help her the most, according to the posted price and her risk-tolerance level. The key challenge lies in that the arm-pulling and experience-trading decisions affect each other. To this end, we design the transaction-based upper confidence bound to estimate the prior-unknown rewards of arms, based on which the agents pull arms or trade their experience. We prove the regret bound of the proposed algorithm for each independent agent and conduct extensive experiments to verify the performance of our solution.

Sparse Mobile Crowdsensing is an emerging paradigm for data collection with budgets and workers' limitations' which recruits workers to sense a part of spatio-temporal data and infer what is unsensed. In order to achieve high inferring accuracy in all spatio-temporal areas, we need to measure the importance level of each area and sense some important ones. Existing works usually use the average distance or the difficulty level inferred by historical data to measure the area's importance. However, we argue that neither distance nor difficulty level is suitable for measuring the importance. A better approach is inspired by the data itself, i.e., data similarity among different areas. Furthermore, there usually exist multiple data types in sparse mobile crowdsensing, which leads to a more complex inference from two-dimensional data (spatial and temporal) to three-dimensional data (spatial, temporal, and data type). In this paper, we study worker recruitment in a multi-task scenario, which aims to recruit workers to sense important data for a three-dimensional inference. Specifically, we first present the SWDTW method to calculate data similarity, which is used to evaluate data importance. Based on this, we further propose an evaluation model for three-dimensional data similarity and measure the importance of each area. Finally, inspired by generalized greedy and simulated annealing, we propose a worker recruitment method named WRGSA, the target of which is selecting workers to sense important areas to enhance the inference accuracy. Extensive experiments have been conducted over three real-world datasets with multiple data types, which can verify the effectiveness of our proposed methods.

Transfer learning refers to transferring the knowledge of a specific domain to a related domain. In cases where the source and the target learner have similar distribution and parameters, transfer learning can reduce the cost of learning and the construction of the target learner and improve the performance of the target learner. In wireless ad-hoc networks, the users connect to networks based on the service location, and various network channels with different levels of quality-of-service (QoS) are available. The wireless channels represent specific ranges of radio frequencies. When the users move from one location to another, the mobile application may switch channels for good quality of service. This paper predicts the wireless channel based on the user's location. Since channel prediction based on location is feasible in one city, the knowledge of channel prediction in one city can be transferred to another city. Thus, transfer learning is applicable and effective in such applications. The paper uses two cities' wireless mapping datasets to predict network channels and uses transfer learning to predict one city's network channels based on the other city's model. Experiments using different initial learning rates during training and different source and target domain data ratios show that transfer learning is feasible for network prediction among different cities.

Software Defined Networks (SDN) have been proposed as a possible development for next generation networking technology. In a SDN, Virtual Network Functions (VNFs) are used to replace the functions of traditional middleboxes. All of these VNFs can be controlled from a centralized controller, which comes with its own security concerns. However, there are many benefits that come from having a centralized controller as traffic can be easily controlled from a single point. This makes it interesting to further study security when it concerns SDN. This work looks to test intrusion detection system (IDS) performance under different configurations in a SDN in order to make security in SDN more robust. In this work, we take two IDSs, Snort and Suricata, and test their performance under different configurations and traffic loads. We test four different configurations: single, chain, parallel, and cross. Our findings seem to suggest that the cross configuration has the best performance of these IDS configurations.

The great achievement of IoT and the wide use of edge devices have brought explosive growth in data. The quality and scale of data determine the performances of machine learning models. Federated learning has attracted widespread attention for its ability to use isolated data and protect data privacy. Models can represent excellent generalization capabilities through federated training. However, the large number of devices and complex models involved in federated training exacerbate the communication costs and degrade the performance of the global model. Although existing approaches can reduce communication costs, they ignore the degradation of global model accuracy in a heterogeneous environment. To alleviate the huge communication costs in federated learning, this paper focuses on reducing upstream and downstream communication frequency while ensuring global model accuracy. We propose a Two-way Delayed Updates method with Model Similarity in Communication-Efficient Federated Learning (FedTDMS). FedTDMS employs personalized local computation to improve global model accuracy on heterogeneous data. Combining 10-cal update relevance check and global model compensation, FedTDMS reduces the communication frequency in Federated Learning. We conduct experiments on the MNIST-FL and CFAR-10-FL datasets. Results show that FedTDMS can greatly optimize communication efficiency while maintaining good global model accuracy.

In federated learning with differential privacy, an obvious phenomenon of local gradient sparsification emerges in some training rounds. When training with low privacy budgets, there is a risk of excessive noise being introduced into the uploaded gradients, leading to a significant decrease in the accuracy of the global model. To tackle the trade-off between privacy protection and model accuracy with low privacy budgets, we propose a differential privacy federated aggregation method based on gradient sparsification (DP-FedAGS), which not only prevents excessive noise addition by protecting only significant gradients, but also accelerates global model convergence by dynamically calculating the weight of the gradient. Experimental results indicate that DP-FedAGS achieves comparable privacy protection to DP-FedAvg and cpSGD, while outperforming DPFedSNLC. Moreover, our approach respectively attains an approximate average test accuracy improvement of $2 .45 \%, 4 . 79$% and $0 . 29$% over the above three methods, rendering DP-FedAGS a promising approach for exploring a balance between privacy protection and model accuracy.

Payment Channel Networks (PCNs) have been introduced as a viable solution to the scalability problem of the popular blockchain. In PCNs, a payment channel allows its end nodes to pay each other without publishing every transaction to the blockchain. A transaction can be routed in the network if there is a path of channels with sufficient funds, and the intermediate routing nodes can ask the transaction sender for a compensatory fee. However, a channel may eventually become depleted and cannot support further payments in a certain direction, as transaction flows from that direction is heavier than flows from the other direction. In this paper, we discuss a PCN node’s possible roles and objectives, and analyze the strategies nodes should take under different roles by considering nodes’ benefits and the network’s performance. Then, we examine two basic network structures (ring and chord) and determine the constraints under which they constitute a Nash equilibrium. Based on the theoretical results, we propose a balance-aware fee-incentivized routing algorithm to guarantee cost-efficient routing, fair fee charging, and the network’s long lasting good performance in general PCNs. Testbed-based evaluation is conducted to validate our theoretical results and to show the feasibility of our proposed approach.

Unattended wireless sensor networks (UWSNs), in which mobile sinks (MSs) are responsible for the data collection, and which are vulnerable to multiple attacks. For example, the adversaries can initiate MS replication attack after compromising a large number of sensor nodes (SNs). To resist such attacks, some scholars have proposed some schemes. In these schemes, it is assumed that MSs are equipped with tamper resistance hardware, so they are pre-distributed most of keys of a key pool for achieving authentication with SNs. However, in outdoor and even sensitive areas, tamper-resistant hardware is not always absolutely safe. Once MSs are compromised, networks become insecure. In this paper, we propose a secure model. The model contains three types of nodes, namely SNs, MSs, and a base station (BS). MSs are responsible for collecting the data encrypted by SNs and forwarding it to BS; BS is responsible for decrypting and analyzing the collected data. During the data collection process, an MS can collect an SN's data only after being authenticated by the SN using the pre-distribution key information. But it cannot decrypt the collected data. In this model, the adversaries can induce a new type of false data injection attack. In the attack, the replicated MSs impersonate uncompromised SNs to send false data to BS by using the compromised key information. If BS accepts a large amount of false data, it will make a wrong judgment. Analysis and simulation show that the proposed model has excellent resistance against MS replication attacks and false data injection attacks by setting appropriate parameter values.

WiFi channel state information (CSI)-based activity recognition has initiated a great many studies because of wide availability and privacy protection. However, general recognition approaches still struggle to generalize beyond the source domain of training data, i.e., well-trained models might not be suitable to target data with unseen subjects or environments. Existing solutions, such as few-shot learning-based and data augmentation-based approaches, either require a few labeled target samples which is difficult to be collected, especially, for old target users, or inappropriately treat augmented samples with different amounts of noise. To overcome these limitations, we propose a Mean Teacher-based cross-domain human activity recognition framework using WiFi CSI, WiTeacher. In this framework, to address the shift between source and target domains, we built a label smoothing-based classification loss, where the input data are the target-like samples generated by StyleGAN, and corresponding label values are dynamically adjusted by our designed adaptive label smoothing method. To enhance the model robustness, we devise a sample relation-based consistency regularization term to keep the distances of the two samples with and without perturbations invariant, which can exploit the relationships between samples to improve recognition performance. The experiments illustrate that WiTeacher achieves obvious gains without requiring any annotation data from the target domain.

Abstract Fog computing is an emerging paradigm that supplies storage, computation, and networking resources between traditional cloud data centers and end devices. This article focuses on the resource provisioning problem in collaborative fog computing for multiple delay‐sensitive users. Our goal is to implement a resource provisioning strategy for network operators to minimize the total monetary cost by considering the deadline and capacity constraints. Two scenarios are considered: unlimited‐processor fog nodes (UPFN) and limited‐processor fog nodes (LPFN). In either scenario, we prove that the resource provisioning problem is NP‐hard. First, we consider the UPFN scenario that the processors of fog nodes are unlimited and users' requests can be ideally processed in parallel. Two algorithms are proposed which greedily delete fog nodes based on the local or global collaborative influences until there is no feasible provisioning to guarantee the deadline of users. Then we extend the resource provisioning problem to a more realistic and complicated scenario LPFN in which the scheduling delay cannot be ignored. Two types of tasks are considered. One is the arbitrarily divided tasks, and a near‐optimal solution bounded by has been found. m is the number of fog nodes, and is the upper bound on the Lipschitz constant of the delay function. Another one is the application‐driven tasks, and we propose a heuristic algorithm. Extensive experiments validate the efficiency of the proposed algorithms.

For the efficient analysis of geo-distributed datasets, cloud providers implement data-parallel jobs across geo-distributed sites (e.g., datacenters and edge clusters), which are generally interconnected by wide-area network links. However, current state-of-the-art geo-distributed data analytic methods fail to make full use of the available network and computing resources. The main reason is that such geo-distributed methods must wait for bottleneck sites to complete the corresponding transmission and computation in each phase. Furthermore, such geo-distributed methods may be impractical to the network bandwidth dynamicity and diverse job parallelism. To this end, we propose a Simultaneous Data Transfer and Processing (SDTP) mechanism to accelerate wide-area data analytics, with the joint consideration of network bandwidth dynamics and job parallelism. In the SDTP, a site can execute the computation, provided that it obtains the required input data. As a result, the input data loading, map, shuffle, and reduce phases at each site need not wait for the completion of the previous phases of other sites. We further improve the SDTP method by offering more accurate time estimation and generalizing the mechanism to dynamic situations. The trace-driven results demonstrate that SDTP can improve the wide-area analytic job response time by 19% to 72% compared to other methods.

Due to the explosive increase of enterprise network traffic, middleboxes that inspect packets through customized rules have been widely outsourced for cost-saving. Despite promising, redirecting enterprise traffic to remote middleboxes raises privacy concerns about the exposure of corporate secrets. To address this, existing solutions mainly apply searchable encryption (SE) to encrypt traffic and rules, enabling middlebox to perform pattern matching over ciphertexts without learning any sensitive information. However, SE is designed for searching pre-chosen keywords, and may cause extensive costs when applied directly to inspecting traffic in which the keywords cannot be determined in advance. The inefficiency of existing SE-based approaches motivates us to investigate a privacy-preserving and lightweight middlebox. To this end, this paper designs$\mathsf{SlimBox}$, which rapidly screens out potentially malicious packets in constant time while incurring only moderate communication overhead. Our main idea is to fragment a traffic/rule string into sub-patterns to achieve conjunctive sub-pattern matching over ciphertexts, while incorporating the position information into the secure matching process to avoid false positives. Experiment results on real datasets show that$\mathsf{SlimBox}$can achieve a good tradeoff between matching latency and communication cost compared to prior work.

Spatial Crowdsourcing (SC) has been adopted in various applications such as Gigwalk and Uber, where a platform takes location-based tasks (e.g., picking up passengers) from a requester and selects suitable workers to perform them. In most existing works, the platform selects workers based on the requester and worker information, which suffers from serious privacy issues. Some works have considered privacy issues, but they still suffer from either of two limitations: (i) Privacy of the requester and worker cannot be protected simultaneously; (ii) Third-party trusted entities are usually required. Motivated by this, we focus on protecting the privacy of both the requester and the worker without third-party entities while selecting workers. We use randomized response, a widely recognized and prevalent privacy model achieving Local Differential Privacy (LDP), to jointly protect the privacy of workers’ locations and charges based on the location-charge correlation. For the requester, we present a novel mechanism called randomized matrix multiplication to hide the real task locations. More importantly, we prove that the worker selection based on the protected information is non-submodular and NP-hard, which cannot be addressed in polynomial time. To this end, we present an approximate algorithm to solve the problem efficiently, of which the effectiveness is measured by the approximation ratio, i.e., the ratio of the optimal solution to the approximate solution. Finally, simulations based on real-world datasets illustrate that our worker selection outperforms the state-of-the-art method on both privacy protection and worker selection.

Points-of-interest (POIs) recommendation plays a vital role in location-based social networks (LBSNs) by introducing unexplored POIs to consumers and has drawn extensive attention from academia and industry. Existing POI recommender systems usually learn fixed latent vectors to represent both consumers and POIs from historical check-ins and make recommendations under the spatio-temporal constraints. However, we argue that the existing works still suffer from the challenges of explaining consumers’ complicated check-in actions. To this end, we first explore the interpretability of recommendations from the POI aspect, i.e., for a specific POI, its function usually changes over time, so representing a POI with a single fixed latent vector is not sufficient to describe the dynamic nature of POIs. Besides, check-in actions to a POI are also affected by the zone where it is located. In other words, the zone's embedding learned from POI distributions, road segments, and historical check-ins could be jointly utilized to enhance POI embeddings. Along this line, we propose aTime-zone-spacePOI embedding model (ToP), which integrates multi-knowledge graphs and topic model to introduce not only spatio-temporal effects but also sentiment constraints into POI embeddings for strengthening interpretability of recommendation. Specifically, ToP learns multiple latent vectors for a POI in a different period with spatial constraints via knowledge graph learning. To add sentiment constraints, ToP jointly combines these vectors with the zone's representations learned by topic models to make explainable recommendations. ToP considers the time, space, and sentiment of POI in a unified embedding framework, which benefits the POI recommendations. Extensive experiments on real-world Changchun city datasets demonstrate that ToP achieves state-of-the-art performance in terms of common metrics and provides more insights for consumers’ POI check-in actions.

The multi-armed bandit (MAB) models have always received lots of attention from multiple research communities due to their broad application domains. The optimal selection problem with unknown rewards in advance, such as ad recommendation in social networks, spectrum access in the cognitive radio field, etc., can be efficiently solved by using MAB models. In an MAB model, given$N$arms whose rewards are unknown in advance, the player selects exactly one arm in each round, and his goal is to maximize the cumulative rewards over a fixed horizon. Further, a more general model called combinatorial MAB (i.e., CMAB), where$K$arms can be played simultaneously in each round, is put forward. However, the existing CMAB models neglect the strategic behaviors of the$N$arms, which indicates that one arm might report false information to increase its own profits. In fact, in many applications such as user selection in crowdsensing, the arms are not the feelingless machines but the rational individuals. To this end, we combine the upper confidence bound (UCB) with auction theory to develop a new algorithm called auction-based UCB (AUCB). We divide the auction-based CMAB problem into two sub-problems: winning arm selection and payment computation problems. For AUCB, we derive an upper bound on regret and prove the truthfulness in one round, individual rationality, and computational efficiency. In addition, we consider an extended situation that some arms may be unavailable in some rounds and the arms will bid inconsistently in different rounds. We devise another algorithm called eAUCB to solve this problem. Extensive simulations are conducted to show the significant performance of the proposed algorithms.

Mobile CrowdSensing (MCS) has recently become a promising data acquisition paradigm, which recruits a large number of users to collect data from the target sensing areas. Obviously, with the increase of sensing scale and the decrease of sensing granularity, traditional MCS cannot fully cover the required sensing areas especially the inaccessible areas. As a variant, Sparse MCS can utilize the spatiotemporal correlations in sensing data to infer the whole sensing map only by sensing a few subareas. However, in many real-world scenarios, such as traffic congestion prediction or parking occupancy detection, inferring the current unsensed data may not be the final goal. By comparison, it is more important to get the future information through the sparse sensed data. In this paper, we turn attention from inferring the current unsensed data to predicting the future unknown data and propose an urban inference and prediction framework in Sparse MCS. To deal with the sparse sensed data, we first present a bipartite-graph-based matrix completion algorithm with spatiotemporal constraints to accurately recover the current full map. Then, by exploiting spatiotemporal correlations based on the inferred full map, we present a Graph Convolutional Networks (GCN) with spatiotemporal attention to predict the future maps. Furthermore, we design a spatiotemporal iterative method to repeatedly update the spatiotemporal attentions and constraints, in order to connect the urban inference and prediction to improve the accuracy of the whole framework. Extensive experiments have been conducted on two types of typical urban sensing tasks with four real-world data sets, which verify the effectiveness of our proposed algorithms in improving the inference and prediction accuracy with the sparse sensed data.

Mobile Crowdsensing (MCS) is a promising paradigm that recruits users to cooperatively perform a sensing task. When recruiting users, existing works mainly focus on selecting a group of users with the best objective ability, e.g., the user's probability or frequency of covering the task locations. However, we argue that the task completion effect depends not only on the user's objective ability, but also on their subjective collaboration likelihood with each other. Furthermore, even though we can find a well-behaved group of users in the single-round scenario, while in the multi-round scenario without enough prior knowledge, we still face the problem of recruiting previously well-behaved user groups (exploitation) or recruiting uncertain user groups (exploration). Additionally, we consider that each user has a different cost, and the platform recruits users under a cost budget; thus, the problem becomes more challenging: users may report fake costs to gain more profits. To address these problems, assuming that the user's information is known, we first convert the single-round user recruitment problem into the min-cut problem and propose a graph theory based algorithm to find the approximate solution. Then, in the multi-round scenario where the user's information is estimated from the previous rounds, to balance the trade-off between exploration and exploitation, we propose the multi-round User Recruitment strategy under the budget constraint based on the combinatorial Multi-armed Bandit model (URMB), which is proven to achieve a tight regret bound. Next, we propose a graph-based payment strategy to achieve truthfulness and individual rationality of users. Finally, extensive experiments on three real-world datasets show that URMB always outperforms the state-of-th-art strategies.

Extracting valuable activity segments from continuously received sensor data is a key step for many downstream applications such as activity recognition, trajectory prediction, and gesture recognition. Numerous unsupervised and supervised approaches have been proposed for activity segmentation. However, current unsupervised methods generally suffer from subject and environment-dependent problems, and supervised methods require a great many labeled data which is time-consuming and expensive to be collected. To address these issues, we propose a Self-supervised Few-shot Time-series Segmentation framework called SFTSeg, which introduces few-shot learning to conduct activity segmentation only relying on several labeled target samples. To be applicable to time-series data, we design a line-level data augmentation method to build a consistency regularization for the few-shot learning framework, which can augment limited labeled target samples to enhance generalization capacity of the model. Also, we devise a time series-specific pretext task to construct a self-supervised loss with adaptive weighting, which can adopt unlabeled target data to enable the model to learn characteristics of the target data and further improve segmentation performance. The experiments illustrated that SFTSeg achieves obvious gains compared to state-of-the-art methods.

In this paper, we investigate the incentive problem in Spatial Crowdsourcing (SC), where mobile social-aware workers have unknown qualities and can share their answers to tasks via social networks. The objectives are to recruit high-quality workers and maximize all parties’ utilities simultaneously. However, most existing works assume that the qualities of workers are known in advance or cannot take all parties’ utilities into account together, especially having not considered the impact of social networks. Thus, we propose an incentive mechanism based on the multi-armed bandit and three-stage Stackelberg game, called TACT. We first design a greedy arm-pulling scheme to recruit workers, which not only can solve the exploration-exploitation dilemma but also takes workers’ social relations into account. Based on the recruitment results, we further design the utility functions incorporating with social benefits for workers, and model the payment computation problem as a three-stage Stackelberg game among all participants. Next, we derive the optimal strategy group so that each party can maximize its own utility to form a multi-win situation. Moreover, we theoretically prove the unique existence of Stackelberg equilibrium and the worst regret bound. Finally, we conduct extensive simulations on a real trace to corroborate the performance of TACT.

With the dawn of 5G network, a new set of requirements for site spectrum monitoring, location-based services (LBS), network construction, and cellular planning are emerging, all of which are relying on fine-grained signal map. Although with significant importance, the traditional signal map construction, e.g., through full site survey, could be time-consuming and labor-intensive as the signal varies frequently over time and the accuracy requirement grows rapidly with the emergence of new applications. The state-of-the arts usually employ crowdsourcing scheme and matrix completion algorithm to solve the dilemma. However, the crowdsourcing scheme usually suffers from uneven distributed and inadequate participants, while the matrix completion methods do not take the specific signal map features into account, thus suffering from sub-optimal recovery results. To this end, in this paper, we study how to effectively reconstruct and update the signal map in the case of partially measured signal maps with smaller cost and propose a GAN-based active signal map reconstruction method (GSMAC). Our method is mainly innovative in two parts: GSMC, GAN-based signal map construction, and ACS, an active crowdsourcing scheme. Specifically, GSMC can effectively update the signal map with only a small number of observations while also fully using the incomplete historical signals to effectively update the signal map online. Meanwhile, ACS consists of a reinforce learning-based active query mechanism which quantitatively evaluates the most valuable measurement site for reconstruction, which further reduces the measurement cost to minimum. The simulation results and real implemented data driven experiments demonstrate the advantages and effectiveness of our approach in both accuracy and cost.

This article proposes a novel Reverse Auction-based Computation Offloading and Resource Allocation Mechanism, named RACORAM for the mobile Cloud-Edge computing. The basic idea is that the Cloud Service Center (CSC) recruits edge server owners to replace it to accommodate offloaded computation from nearby resource-constraint Mobile Devices (MDs). In RACORAM, the reverse auction is used to stimulate edge server owners to participate in the offloading process, and the reverse auction-based computation offloading and resource allocation problem is formulated as a Mixed Integer Nonlinear Programming (MINLP) problem, aiming to minimize the cost of the CSC. The original problem is decomposed into an equivalent master problem and subproblem, and low-complexity algorithms are proposed to solve the related optimization problems. Specifically, a Constrained Gradient Descent Allocation Method (CGDAM) is first proposed to determine the computation resource allocation strategy, and then a Greedy Randomized Adaptive Search Procedure based Winning Bid Scheduling Method (GWBSM) is proposed to determine the computation offloading strategy. Meanwhile, the CSC's payment determination for the winning edge server owners is also presented. Simulations are conducted to evaluate the performance of RACORAM, and the results show that RACORAM is very close to the optimal method with significantly reduced computational complexity, and greatly outperforms the other baseline methods in terms of the CSC's cost under different scenarios.

The Point Of Interest (POI) sequence recommendation is the key task in itinerary and travel route planning. Existing works usually consider the temporal and spatial factors in travel planning. However, the external environment, such as the weather, is usually overlooked. In fact, the weather is an important factor because it can affect a user’s check-in behaviors. Furthermore, most of the existing research is based on a static environment for POI sequence recommendation. While the external environment (e.g., the weather) may change during travel, it is difficult for existing works to adjust the POI sequence in time. What’s more, people usually prefer the attractive routes when traveling. To address these issues, we first conduct comprehensive data analysis on two real-world check-in datasets to study the effects of weather and time, as well as the features of the POI sequence. Based on this, we propose a model of Dynamic Personalized POI Sequence Recommendation with fine-grained contexts ( DPSR for short). It extracts user interest and POI popularity with fine-grained contexts and captures the attractiveness of the POI sequence. Next, we apply the Monte Carlo Tree Search model (MCTS for short) to simulate the process of recommending POI sequence in the dynamic environment, i.e., the weather and time change after visiting a POI. What’s more, we consider different speeds to reflect the fact that people may take different transportation to transfer between POIs. To validate the efficacy of DPSR , we conduct extensive experiments. The results show that our model can improve the accuracy of the recommendation significantly. Furthermore, it can better meet user preferences and enhance experiences.

Bitcoin builds upon an unstructured peer-to-peer overlay network to disseminate transactions and blocks. Broadcast in such a network is slow and brings inconsistencies, i. e., peers have different views of the system state. Due to the delayed block propagation and the competition of mining, forking, i. e., the blockchain temporarily diverges into two or more branches, occurs, which wastes computation power and causes security issues. This paper proposes an autonomous and distributed topology optimization mechanism to reduce block propagation delay and hence reduce the occurrence of blockchain forks. In the proposed mechanism, a node can autonomously update his neighbor set using the information provided by his current neighbors, since each neighbor will recommend a peer from his own neighbor set, i. e., a neighbor’s neighbor, to this node. Each recommendation is based on a peer’s propagation ability, which is characterized as a criteria function obtained through a combination of empirical analysis and machine learning. We further propose some metrics to evaluate a Bitcoin network topology. Experiment results reflect the effectiveness of the proposed mechanism and indicate the correlation between block propagation time and fork rate. Thus, we analyze the relation between block propagation time and fork rate by applying an epidemic model to capture the block propagation process. We prove that a Bitcoin network topology with a relatively small network delay variance among all nodes produces a lower fork rate than another topology if its average block propagation time to 84% of the entire network is shorter.

Mobile CrowdSensing (MCS) is a popular data collection paradigm which usually faces the problem of sparse sensed data because of the limited sensing cost. In order to address the situation of sparse data, sparse MCS recruits users to sense important areas and infers completed data by data completion, which is crucial in sparse MCS for urban sensing applications (e.g. enhancing data expression, improving urban analysis, guiding city planning, etc.) To achieve accurate completion results, previous methods usually utilize the universal similarity and conventional tendency while incorporating only a single dataset to infer the full map. However, in real-world scenarios, there may exist many kinds of data (inter-data), that could help to complement each other. Moreover, for each kind of data (intra-data), there usually exist a few but important outliers caused by the special events (e.g., parking peak, traffic congestion, or festival parade), which may behave in a different way as the statistical patterns. These outliers cannot be ignored, while it is difficult to detect and recover them in data completion because of the following challenges: 1) the infrequency and unpredictability of outliers’ occurrence, 2) the large deviations against the means compared to normal values, and 3) the complex spatiotemporal relations among inter-data. To this end, focusing on spatiotemporal data with both intra- and inter-data correlations, we propose a matrix completion method that takes outliers’ effects into consideration and exploits both intra- and inter-data correlations for enhancing performance. Specifically, we first conduct the Deep Matrix Factorization (DMF) with multiple auxiliary Neural Networks, which named Stacked Deep Matrix Factorization (SDMF). Note that the loss function of SDMF is no longer the previous MSE loss function, but replaced with an Outlier Value Loss (OVL) function to effectively detect and recover the outliers. Moreover, a spatiotemporal outlier value memory network is added for further enhancing the outlier inference. Finally, we take extensive qualitative and quantitative experiments on two popular datasets each with two types of sensing data, and the experimental results indicate the advantages of our method that outperforms the state-of-the-art methods.

Mobile edge computing (MEC) reduces data service latency by pushing data to the network edge. However, due to the dynamic and diverse requests of mobile users, the problem of mobile edge caching is more complex than cloud caching. Therefore, the existing model-based caching strategies cannot be directly used in the mobile edge caching environment. Besides, when taking the cooperative storage relationship between neighbor edge servers into consideration, the caching problem becomes more difficult. To this end, we formulate an mobile edge caching problem to minimize the total latency in mobile edge computing. Firstly, a heuristic caching strategy is proposed to solve the mobile edge caching problem in the single-time-slot scenario. Then, with the consideration of users’ mobility and the correlation of files, we propose a caching strategy for the multiple-time-slot scenario based on multi-agent deep reinforcement learning. To address the cold start problem in deep reinforcement learning, we adopt the proposed heuristic caching strategy used in the single-time-slot scenario to further optimize the training results. Extensive experiments on generated data and real-world datasets are conducted to verify that the proposed edge caching strategies can achieve the minimum latency compared with the state-of-the-art strategies.

Loosely coupled and highly cohesived microservices running in containers are becoming the new paradigm for application development. Compared with monolithic applications, applications built on microservices architecture can be deployed and scaled independently, which promises to simplify software development and operation. However, the dramatic increase in the scale of microservices and east-west network traffic in the data center have made the cluster management more complex. Not only does the scale of microservices cause a great deal of pressure on cluster management, but also cascading QoS violations present a substantial risk for SLOs (Service Level Objectives). In this paper, we propose a Microservice-Oriented Topology-Aware Scheduling Framework (MOTAS), which effectively utilizes the topologies of microservices and clusters to optimize the network overhead of microservice applications through a heuristic graph mapping algorithm. The proposed framework can also guarantee the cluster resource utilization. To deal with the dynamic environment of microservice, we propose a mechanism based on distributed trace analysis to detect and handle QoS violations in microservice applications. Through real-world experiments, the framework has been proved to be effective in ensuring cluster resource utilization, reducing application end-to-end latency, improving throughput, and handling QoS violations.

With the rapid development of Internet-of-Things (IoT) and the explosive advance of deep learning, there is an urgent need to enable deep learning inference on IoT devices in Mobile Edge Computing (MEC). To address the computation limitation of IoT devices in processing complex Deep Neural Networks (DNNs), computation offloading is proposed as a promising approach. Recently, partial computation offloading is developed to dynamically adjust task assignment strategy in different channel conditions for better performance. In this paper, we take advantage of intrinsic DNN computation characteristics and propose a novel Fused-Layer-based (FL-based) DNN model parallelism method to accelerate inference. The key idea is that a DNN layer can be converted to several smaller layers in order to increase partial computation offloading flexibility, and thus further create the better computation offloading solution. However, there is a trade-off between computation offloading flexibility as well as model parallelism overhead. Then, we investigate the optimal DNN model parallelism and the corresponding scheduling and offloading strategies in partial computation offloading. In particular, we propose a Particle Swarm Optimization with Minimizing Waiting (PSOMW) method, which explores and updates the FL strategy, path scheduling strategy, and path offloading strategy to reduce time complexity and avoid invalid solutions. Finally, we validate the effectiveness of the proposed method in commonly used DNNs. The results show that the proposed method can reduce the DNN inference time by an average of 12.75 times compared to the legacy No FL (NFL) algorithm, and is very close to the optimal solution achieved by the Brute Force (BF) algorithm with the difference of less than 0.04%.

Blockchain came to prominence as the distributed ledger underneath Bitcoin, which protects the transaction histories in a fully-connected, peer-to-peer network. The blockchain mining process requires high computing power to solve a Proof-of-Work (PoW) puzzle, which is hard to implement on users’ mobile devices. So these miners may leverage the edge/cloud service providers (ESPs/CSP) to calculate the PoW puzzle. The existing edge-assisted blockchain networks assumed that all ESPs have a uniform propagation delay, which is unrealistic. In this article, we consider a more practical scene where ESPs locate in diverse positions of the blockchain network, which causes different propagation delays when supporting the computation of the PoW puzzle. Additionally, these ESPs connect to a remote CSP for resource scheduling when the computing tasks exceed their maximum capacity. The blockchain mining process generally involves complicated competition and games among CSP, ESPs, and miners. Each service provider focuses on how to determine his resource price so that he can maximize his utility. According to the set resource price, each miner concentrates on scheduling his resource requests for each ESP to maximize individual personal utility, which depends on ESPs’ resource price and propagation delays. We first model such a resource pricing and scheduling problem as a three-stage multi-leader multi-follower Stackelberg game and aim at finding the Stackelberg equilibrium. Then, we analyze the subgame optimization problem in each stage and propose an iterative algorithm based on backward induction to achieve the Nash equilibrium of the Stackelberg game. Finally, extensive simulations are conducted to verify the significant performance of the proposed solution.

Edge caching is a promising approach to reduce duplicate content transmission in Internet-of-Vehicles (IoVs). Several Reinforcement Learning (RL) based edge caching methods have been proposed to improve the resource utilization and reduce the backhaul traffic load. However, they only obtain the local sub-optimal solution, as they neglect the influence from environments by other agents. This paper investigates the edge caching strategies with consideration of the content delivery and cache replacement by exploiting the distributed Multi-Agent Reinforcement Learning (MARL). A hierarchical edge caching architecture for IoVs is proposed and the corresponding problem is formulated with the goal to minimize the long-term content access cost in the system. Then, we extend the Markov Decision Process (MDP) in the single agent RL to the context of a multi-agent system, and tackle the corresponding combinatorial multi-armed bandit problem based on the framework of a stochastic game. Specifically, we firstly propose a Distributed MARL-based Edge caching method (DMRE), where each agent can adaptively learn its best behaviour in conjunction with other agents for intelligent caching. Meanwhile, we attempt to reduce the computation complexity of DMRE by parameter approximation, which legitimately simplifies the training targets. However, DMRE is enabled to represent and update the parameter by creating a lookup table, essentially a tabular-based method, which generally performs inefficiently in large-scale scenarios. To circumvent the issue and make more expressive parametric models, we incorporate the technical advantage of the Deep-$Q$Network into DMRE, and further develop a computationally efficient method (DeepDMRE) with neural network-based Nash equilibria approximation. Extensive simulations are conducted to verify the effectiveness of the proposed methods. Especially, DeepDMRE outperforms DMRE,$Q$-learning, LFU, and LRU, and the edge hit rate is improved by roughly 5%, 19%, 40%, and 35%, respectively, when the cache capacity reaches 1, 000 MB.

Sizes of data processing jobs in cloud clusters have been growing rapidly in the big data era. It is critical to execute those jobs efficiently. The average job completion time (JCT) is a widely used metric to measure executing efficiency. JCT refers to the length of the time interval between a job's arrival to its completion. Typically, a data processing job contains multiple stages with complex precedence constraints. Carefully scheduling the processing sequence of stages within a job may significantly reduce its JCT. Our objective is to minimize the average JCT for online arrival jobs. The computation graphs of those jobs are usually directed acyclic graphs (DAGs). It makes the scheduling problem challenging. Recent works have shown that reinforcement learning (RL) agents can adaptively adjust the scheduling policies by dynamically assigning priorities for job stages. However, we notice that other factors besides stage priories may impact the JCT significantly. In particular, we observe that inserting idle slots before large jobs may reduce the waiting time of small jobs that arrive slightly later and reduce the average JCT. We analyze the benefits of inserting idle time for simple cases theoretically and show the condition in which idle slots should be inserted for two adjacent jobs. In addition, we adapt the RL-based scheduler by integrating the observation. Experiment results on both real-world and synthetic datasets show the efficiency of our scheduler. Also, a perturbation-based method is applied to demonstrate the contribution of each proposed feature.

With the development of Internet of Things (IoT) and the advance of deep learning, there is an urgent need to enable deep learning inference on IoT devices. To address the computation limitation of IoT devices in processing complex Deep Neural Networks (DNNs), partial computation offloading is developed to dynamically adjust computation offloading assignment strategy in different channel conditions for better performance. In this paper, we take advantage of intrinsic DNN computation characteristics, and propose a novel Fused-Layer-based (FL-based) DNN model parallelism method to accelerate inference. The key idea is that a DNN layer can be converted to several smaller layers to increase partial computation offloading flexibility, and thus further create better computation offloading solution. However, there is a trade-off between parallelism computation offloading flexibility and model parallelism overhead. Then, we discuss the optimal DNN model parallelism and the corresponding scheduling and offloading strategies in partial computation offloading. In particular, we present a Minimizing Waiting (MW) method, which explores both the FL strategy, the path scheduling strategy, and the path offloading strategy to reduce time complexity. Finally, we validate the effectiveness of the proposed method in commonly used DNNs. The results show that the proposed method can reduce the DNN inference time by an average of 18.39 times compared with No FL (NFL) algorithm, and is very close to the optimal solution Brute Force (BF) with greatly reduced time complexity.

The concept of competitive influence maximization has attracted considerable research interest in recent years. Modeling the behavior of social network members is a key challenge in this regard. This kind of forwarding occurs from newly activated nodes to their not yet activated neighbors, especially when some competitors try to gain maximum influence over the network. For competitors, the main objective is to find which potential members are most valuable to them and how many resources they should allocate to them in order to get as much influence as possible. Previous studies on competitive influence maximization tended to focus on the single-shot game without considering how budget allocation affects the outcome. We are interested in maximizing the total number of activated nodes through multi-round competitive influence maximization, where each competitor has to determine the locations and budget amount to invest simultaneously and repeatedly within a given total budget. In this paper, we propose a tree-approximate game-theoretical framework and introduce the new measurement as a dynamic weight. This measure allows us to isolate the most influential member with a high degree of accuracy. Our simulation study demonstrates the effectiveness of our approach in solving a multi-round, learning-based CIM problem.

Cyber-security research often focuses on attack-defense games where a strategic attacker seeks to destroy the defender's targets or kill him. In such a game there is a defender who just try to protect himself. In the real world, players can choose to protect themselves as well as kill their opponents to maximize their overall gain. In this case, the player allocates their budget for both defending and attacking actions and decides how well to attack and how well to defend against others. Players should allocate their budgets appropriately for each action throughout multiple rounds when playing such a game. The probabilities of surviving and killing in each round are determined by what happened in the previous rounds and the amount of the remaining budget. Players can continue playing until they die or run out of money. It is possible for the player not to be aware of everything his opponent does. Despite knowing that his opponent is playing according to one of the possible types, he cannot see which action exactly he is taking. Likewise, it may be the case that the player only sees the opponent's action, but does not know what its objective is. To meet this challenge, this paper develops a game where players decide how to allocate resources when they have partial information. For a model with complete information, equilibrium can be found and, as an extension, models with incomplete and imperfect information are also discussed. Our simulation examines how utility changes based on prior beliefs, total budgets, costs, and uncertainty.

As the popularity of mobile devices continues to increase, Mobile Crowdsensing (MC), a scalable and efficient data collection method, has received widespread attention. Although lots of effort has been devoted to studying the task assignment or worker recruitment in MC, most of them focus on how to maximize the profit from the perspective of the platform while ignoring rational individual workers' entitlement. We creatively start from the worker's perspective to find the task selection strategy to maximize the worker's profit. In this paper, the problem of unknown task selection is modeled as a Multi-Armed Bandit (MAB), on which three types of additional constraints are considered. The first constraint is the device budget. Workers choose and conduct tasks before it is exhausted. The second constraint is the personal preference regarding the traveling cost. The third constraint is the balance requirement of the MC platform, which has regulations on the tasks' execution rounds. In addition to the dilemma between exploration and exploitation in the classical MAB, we have to face the tradeoff between the reward and all the constraints above. To this end, we first adopt the epoch-style algorithm to reduce the number of switches between any two sensing tasks and further build new algorithms to deal with different constraints. The traveling cost and platform balance are involved in the task index computation as a penalty. We conduct extensive simulations based on real-world traces to verify the significant performance of our proposed algorithms.

Deep Neural Network (DNN) models have been widely deployed in a variety of applications. Driven by privacy concerns and great improvement in the computational power of mobile devices, the idea of training machine learning models on mobile devices has become more and more important. Directly applying parallel training frameworks designed for data center networks to train DNN models on mobile devices may not achieve the ideal performance, since mobile devices usually have multiple types of computation resources such as ASIC, neural engine, and FPGA. Moreover, the communication time is not negligible when training on mobile devices. With the objective of minimizing DNN training time, we propose to extend the pipeline parallelism, which can hide the communication time behind computation for DNN training by integrating the resource allocation. Fine-tuning the ratio of resources allocated to forward and backward propagation can improve resource utilization. We focus on homogeneous workers and theoretically analyze the ideal cases where resources are linearly separable. We also discuss the model partition and resource allocation for a more realistic case. Additionally, we investigate the heterogeneous worker case. Trace-based simulation results show that our scheme can efficiently reduce the time cost of a training iteration.

Service mesh is a promising micro-services architecture due to its excellent governance capabilities. Unlike traditional service invocation, configurations for governance need to be issued in the service mesh. However, we find that the control-plane traffic of governance is distributed in full by default, i.e., each service in the data plane receives all configurations. The vast majority of the configurations are redundant for a specific service. Hence, it is important and challenging to make the control plane aware of the calling relationships between services. In this paper, we propose a traffic management mechanism named DATM. Using this mechanism, the entire cluster can be dynamically controlled and services can be configured on demand. It is implemented through a dependency-aware controller and monitors. The controller first processes the information listened to by the monitors and then analyzes the connection between the metrics and the service requests through intelligent algorithms. Finally, the control traffic for regulating the control plane is generated. Our proposed mechanism is experimentally compared with the default strategy and existing work across a wide set of load scenarios in a testbed based on Istio service mesh and Kubernetes. Experimental results demonstrate that our mechanism can save the storage resources of a single agent by 40% to 60%, and the number of cluster updates can be greatly reduced. From the perspective of the whole cluster, the optimization results are even better.

With the advent of the Internet of Things (IoT) era and 5G, ubiquitous sensing devices (e.g., smartphones, surveillance sites, and security cameras) have been widely used in various fields, resulting in the generation of a huge amount of monitoring data. The rise of federated learning makes it possible to leverage monitoring data to train deep neural networks through cloud-edge collaboration without compromising privacy. However, the non identically and independently distributed (called Non-IID) data collected by IoT devieces creates a client drift phenomenon, resulting in a slow convergence of the global model. To this end, we propose a new Federated learning framework based on Gradient Variance Reduction with a correction weight control mechanism and Global gradient descent with Momentum, named FedGVRGM to conduct gradient correction and reduce the negative impacts of prediction parameters. Specifically, in the local training phase, FedGVRGM combines gradient variance reduction with a correction weight control mechanism to further correct the local model parameters, thus reducing the dispersion of model parameters among clients. In the global aggregation phase, FedGVRGM integrates the historical change states of the global model through the gradient descent with momentum to reduce the oscillations and improve the convergence speed of the global model. We refer to the above methods of gradient adjustment in the local and global training phases as FedGVR and FedGM, respectively. Numerous evaluations are conducted on CIFAR-100, CIFAR-10, and MNIST datasets to prove that FedGVRGM has a faster convergence rate than other stateof-the-art approaches such as Federated Averaging (FedAvg), FedProx, FedReg, FedGVR, and FedGM.

As a cost-effective paradigm, Sparse Crowdsensing aims to recruit workers to perform a part of sensing tasks and infer the rest, which has broad applications, including environmental monitoring, urban sensing, etc. In most cases, workers will participate in real time, and thus their sensing data are coming dynamically. Taking full advantage of the online coming data to complete the full sensing map is an important problem for Sparse Crowdsensing. However, for data completion, the importance of data collected from different spatio-temporal areas is usually different and time-varying. For example, the newly obtained data in the center is often more important than the old ones from edges. Moreover, the area importance may also influence the following worker selection, i.e., selecting suitable workers to actively sense important areas (instead of passively waiting for given data) for improving completion accuracy. To this end, in this paper, we propose a framework for online Sparse Crowdsensing, called OS-MCS, which consists of three parts: matrix completion, importance estimation, and worker selection. We start from the dynamically coming data and propose an online matrix completion algorithm with spatio-temporal constraints. Based on that, we estimate the spatio-temporal area importance by conducting a reinforcement learning-based up-to-date model. Finally, we utilize the prophet secretary problem to select suitable workers to sense important areas for accurate completion in an online manner. Extensive experiments on real-world data sets show the effectiveness of our proposals.

While cloud-based deep learning benefits for high-accuracy inference, it leads to potential privacy risks when exposing sensitive data to untrusted servers. In this paper, we work on exploring the feasibility of steganography in preserving inference privacy. Specifically, we devise GHOST and GHOST+, two private inference solutions employing steganography to make sensitive images invisible in the inference phase. Motivated by the fact that deep neural networks (DNNs) are inherently vulnerable to adversarial attacks, our main idea is turning this vulnerability into the weapon for data privacy, enabling the DNN to misclassify a stego image into the class of the sensitive image hidden in it. The main difference is that GHOST retrains the DNN into a poisoned network to learn the hidden features of sensitive images, but GHOST+ leverages a generative adversarial network (GAN) to produce adversarial perturbations without altering the DNN. For enhanced privacy and a better computation-communication trade-off, both solutions adopt the edge-cloud collaborative framework. Compared with the previous solutions, this is the first work that successfully integrates steganography and the nature of DNNs to achieve private inference while ensuring high accuracy. Extensive experiments validate that steganography has excellent ability in accuracy-aware privacy protection of deep learning.

Recently, the growing popularity of Spatial Crowd-sourcing (SC), allowing untrusted platforms to obtain a great quantity of information about workers and tasks’ locations, has raised numerous privacy concerns. In this paper, we investigate the privacy-preserving task assignment in the online scenario, where workers and tasks arrive at the platform in real time and tasks should be assigned to workers immediately. Traditional online task assignments usually make a benchmark to decide the following task assignment. However, when location privacy is considered, the benchmark does not work anymore. Hence, how to assign tasks in real time based on workers and tasks’ obfuscated locations is a challenging problem. Especially when many tasks could be assigned to one worker, path planning should be considered, making the assignment more challenging. To this end, we propose a Planar Laplace distribution based Privacy mechanism (PLP) to obfuscate real locations of workers and tasks, where the obfuscation does not change the ranking of these locations’ relative distances. Furthermore, we design a Threshold-based Online task Assignment mechanism (TOA), which could deal with the one-worker-many-tasks assignment and achieve a satisfactory competitive ratio. Simulations based on two real-world datasets show that the proposed algorithm consistently outperforms the state-of-the-art approach.

Edge computing is a new computing paradigm rising gradually in recent years. Applications, such as object detection, virtual reality and intelligent cameras, often leverage Deep Neural Networks (DNN) inference technology. The traditional paradigm of DNN inference based on cloud suffers from high delay because of the limited bandwidth. From the perspective of service providers, caching DNN models on the edge brings several benefits, such as efficiency, privacy, security, etc.. The problem we concerned in this paper is how to decide the cached models and how to allocate processors of edge servers to reduce the overall system cost. To solve it, we model and study the DNN Model Caching and Processor Allocation (DMCPA) problem, which considers user-perceived delay and energy consumption with limited edge resources. We model it as an integer nonlinear programming (INLP) problem, and prove its NP-Completeness. Since it is considered as a long-term average optimization problem, we leverage the Lyapunov framework to develop a novel online algorithm DMCPA-GS-Online with Gibbs Sampling. We give the theoretical analysis to prove that our algorithm is near-optimal. In experiments, we study the performance of our algorithm and compare it with other baselines. The simulation results with the trace dataset from real world demonstrate the effectiveness and adaptiveness of our algorithm.

The vigorous development of IoT technology has spawned a series of applications that are delay-sensitive or resource-intensive. Mobile edge computing is an emerging paradigm which provides services between end devices and traditional cloud data centers to users. However, with the continuously increasing investment of demands, it is nontrivial to maintain a higher quality-of-service (QoS) under the erratic activities of mobile users. In this paper, we investigate the service provisioning and updating problem under the multiple-users scenario by improving the performance of services with long-term cost constraints. We first decouple the original long-term optimization problem into a per-slot deterministic one by using Lyapunov optimization. Then, we propose two service updating decision strategies by considering the trajectory prediction conditions of users. Based on that, we design an online strategy by utilizing the committed horizon control method looking forward to multiple slots predictions. We prove the performance bound of our online strategy theoretically in terms of the trade-off between delay and cost. Extensive experiments demonstrate the superior performance of the proposed algorithm.

Welcome to the 18th International Conference on Mobility, Sensing and Networking (MSN 2022), held in Guangzhou, China, on December 14-16, 2022. MSN provides an open forum for academic researchers and industry practitioners to present their research progresses, exchange new ideas, and identify future directions in the fields of mobility, sensing and networking.

In this paper, we address the characterization of similarity between two storage traces and define a three-part measure called SIST (Similarity Index for Storage Traffic). Such a measure is essential for identifying traces that are most appropriate for storage system evaluations. We compare SIST against several other similarity measures in the literature on both the object storage and block storage systems, and show the superiority of SIST in terms of its behavior for known perturbations.

Summary In recent times, many location‐based service providers (LBSPs) choose to outsource data query services to third‐party cloud service providers (CSPs). This allows users to easily search for points of interests (POIs), such as restaurants and parking lots in their vicinity, using their mobile devices and in‐vehicle infotainment units. Skyline query is one potential technique to be deployed for road networks. However, the untrusted CSPs may forge or omit query results, intentionally or not. Therefore, in this article, we posit that by observing the unique properties of skyline query results in road networks, we can bind each POI with four nearby POIs with special properties using signature chain technology. Our proposed approach not only provides users with skyline query result authentication ability over the road network, but also have low communication overhead. Specifically, the overhead analysis and experimental results show that our proposed approach decreases the communication overhead.

In edge networks, distributed computing resources have been widely utilized to collaboratively perform a machine learning task by multiple nodes. However, the model training time in heterogeneous edge networks is becoming longer because of excessive computation and delay caused by slow nodes, namely, stragglers. The parameter server even abandons stragglers which fail to return the outcome within a reasonable deadline, called straggler dropout, decreasing the model accuracy. To optimize the computation cost and maintain the model accuracy, we focus on mitigating the heavy computation of stragglers and preventing straggler dropout. Therefore, we propose a novel scheme named dynamic grouping and heterogeneity-aware gradient coding (DGH-GC) to tolerate stragglers by employing dynamic grouping and gradient coding. DGH-GC evenly distributes stragglers in each group and encodes gradients based on their computation capacity to prevent them drop out. However, DGH-GC exacerbates the communication burden by making data duplication to tolerate stragglers. Relying on the scheme, we further propose an algorithm called DGH-(GC)2 to compress transferred gradients in both upstream communication and downstream communication. Experimental evaluations prove that DGH-(GC) outperforms all state-of-the-art methods and DGH-(GC)2 further speeds up the convergence time of the trained model and saves about 26% average iteration time compared to the DGH-(GC).

Traditional spatial crowdsourcing (SC) systems employ a centralized server platform to provide services for requesters. Such a centralized design requires powerful resource capacity and often cannot accomplish the urgent demands due to the unpredictable network latency. In order to ensure the scalability of systems and the quality of services, we study the distributed SC (DSC), where a diversity of location-relative services provided by various service providers (SPs) can deploy on edge clouds (ECs) with low time latency. Since the edge resources are limited, SPs need to compete for edge resources so as to deploy their desired SC services, and the requested resources must be allocated together to meet the demand of the service. We first design a gated recurrent unit with particle filter (GRUPF) network for SPs to predict future resource demands so as to participate in the competitions judiciously. Then, we model the competitive edge resource allocation problem between SPs and ECs as a combinatorial auction process. Due to the NP-hardness of this problem, an approximation algorithm is proposed to tackle it. Moreover, the leakage of private information such as bids may incur severe economic damage, and most existing studies usually rely on a trusted third party to provide rigorous privacy protection. Therefore, we customize a novel differentially private resource auction (DRA) mechanism, and design a bid confusion strategy based on differential privacy. Through theoretical analysis, we prove that the DRA mechanism meets some desired properties, including$\epsilon $-differential privacy, individual rationality, computational efficiency, and$\gamma $-truthfulness. Additionally, we corroborate the significant performances of DRA through extensive simulations on synthetic and real-world data sets.

With the rapid growth of the coronavirus disease of 2019 (COVID-19) cases, massive amounts of relevant data are being trained on machine learning models for countering communicable infectious diseases. Federated Learning (FL) is a paradigm of distributed machine learning to deal with the individual COVID-19 data, and enable the protection of data privacy. However, FL has low efficiency in Edge-Based wireless communication systems with system heterogeneity. In this paper, we propose an “Asynchronous-Adaptive FL” (AAFL) scheme. Specifically, we allow that medical devices with different performances have a heterogeneous number of local SGD iterations in each communication round, called asynchronous iteration strategy which is balanced under adaptive control. We theoretically analyze the convergence of the AAFL scheme under a given time budget and obtain a mathematical relationship between the heterogeneous number of local SGD iterations and the optimal model parameters. Based on the mathematical relationship, we design an algorithm for parameter server and work nodes to adaptively control the heterogeneous number of local SGD iterations. Subsequently, we build a prototype heterogeneous system and conduct experiments on various scenarios for analyzing the general properties of our algorithm, and then apply our algorithm to public COVID-19 databases. The experimental results and application performance demonstrate the effectiveness and efficiency of our AAFL scheme.

With the development of communications, networking, and information technology, Crowdsensed Data Trading (CDT) becomes a novel data trading paradigm. In CDT, the data requesters publish crowdsensing tasks with specific data requirements, and then workers complete these tasks, upload the data and obtain corresponding rewards. To efficiently deal with data trading, most of the existing CDT systems assume a trusted centralized platform. However, we argue that the platform may collude with workers or requesters to trick others for achieving more benefits. For example, according to the workers’ uploaded data, the platform can modify the reward functions by colluding with the requester. Similarly, the platform might collude with workers to let them know the reward function, then workers could forge data. Meanwhile, requesters and workers may also be malicious. For example, requesters may post tasks but fail to pay and workers can upload wrong data to mislead the system. To solve the above problems, we combine the Crowdsensed Data Trading system with intelligent Blockchain (CDT-B), which contains a smart contract called CDToken. As a credible third-party, the CDToken is used to record the requesters’ reward function and workers’ data uploading function to avoid targeted trick. At the same time, we not only design a Data Uploading and Preprocessing (DUP) mechanism in CDToken to collect and process the workers’ sensed data, but also propose a Grouping Truth Discovery (GTD) to evaluate their data quality for determining the payments. Moreover, to hold a large number of requesters and workers in CDT-B, we propose a Layered Sharding blockchain based on Membership Degree (LSMD) to solve the blockchain inefficiency problem. Finally, we deploy CDToken to an experimental environment based on Ethereum and demonstrate its efficient performance and practicability.

A distributed denial-of-service (DDoS) attack is a cyber-attack in which attackers from different locations send out many requests to exhaust the capacity of a server. Current DDoS attack protection services filter out the DDoS attack packets in the middle of the path from the attacker to the servers. Some of the DDoS protection systems filter them out at the victim server. As a result, unnecessary attack traffic congests the network and wastes bandwidth. This can be minimized if we block them as early as possible. In this paper, we propose a DDoS attack protection system by using the filter router. The victim needs to wisely select and send filters to a subset of filter routers to minimize attack traffic and blockage of legitimate users (LUs). Many filters can easily minimize the attack traffic and blockage of LUs, but it is costly to the victim. So, we formulate two problems with different settings for selecting filter routers given a constraint on the number of filters. We propose dynamic programming solutions for both problems. Both problems consider the blockage of all attack traffic before it reaches the victim. We conduct extensive simulation to support our solutions.

Accessibility to smart devices provides opportunities for side-channel attacks (SCAs) on artificial intelligent (AI) models in the intelligent Internet of Things (IoT). However, the existing literature exposes some shortcomings: 1) incapability of quantifying and analyzing the leaked information through side channels of the intelligent IoT and 2) inability to devise efficient and accurate SCA algorithms. To address these challenges, we propose a side-channel fuzzy analysis-empowered AI model extraction attack in the intelligent IoT. First, the integrated AI model extraction framework is proposed, including power trace-based structure, execution time-based metaparameters, and hierarchical weight extractions. Then, we develop the information theory-based analysis for the AI model extraction via SCA. We derive a mutual information-enabled quantification method, theoretical lower/upper bounds of information leakage, and the minimum number of attack queries to obtain accurate weights. Furthermore, a fuzzy gray correlation-based multiple-microspace parallel SCA algorithm is proposed to extract model weights in the intelligent IoT. Based on the established information-theoretic analysis model, the proposed fuzzy gray correlation-based SCA algorithm obtains high-precision AI weights. Experimental results, consisting of simulation and real-world experiments, verify that the developed analysis method with the information-theoretic perspective is feasible and demonstrate that the designed fuzzy gray correlation-based SCA algorithm is effective for AI model extraction.

Bike-sharing systems usually suffer from out-of-service events due to bike underflow or overflow. We propose to recruit workers to rebalance station loads. We partition the complex rebalancing problem in temporal and spatial domains. The temporal domain is divided into a sequence of slices with a fixed duration. In each slice, we allocate a pair of overflow/underflow stations to a worker such that the cost is minimized, which is NP-hard. A 3-approximation algorithm is proposed. We further investigate the worker shortage case and extend the matching algorithm to consider the number of unsatisfied users. Then, the configuration dynamic in the sequence of slices is captured by determining the rebalancing target for each rebalancing operation. We investigate heuristic approaches to minimize the total number of bike movements. Furthermore, we extend our scheme to dockless BSSs using clustering techniques. We simulate our algorithms on both real-world and synthetic datasets. Experiment results show that our approaches can reduce the average total detour per slice. In worker shortage, considering the number of unsatisfied users could improve the long-term performance of rebalancing. Besides, we find that our scheme could maintain worker satisfaction over multiple time slices, which indicates the sustainability of our rebalancing scheme.

Mobile crowdsensing, through which a requester can coordinate a crowd of workers to complete some sensing tasks, has attracted significant attention recently. In this paper, we focus on the unknown worker recruitment problem in mobile crowdsensing, where workers’ sensing qualities are unknown a priori. We consider the scenario of recruiting workers to complete some continuous sensing tasks. The whole process is divided into multiple rounds. In each round, every task may be covered by more than one recruited workers, but its completion quality only depends on these workers’ maximum sensing quality. Each recruited worker will incur a cost and each task is attached a weight to indicate its importance. Our objective is to determine a recruiting strategy to maximize the total weighted completion quality under a limited budget. We model such unknown worker recruitment process as anovel combinatorial multi-armed bandit(CMAB) problem, and propose an unknown worker recruitment algorithm based on the modified upper confidence bound (UCB). Moreover, we extend the problem to the case where the workers’ costs are also unknown and design the corresponding algorithm. We analyze the regret bounds of the two proposed algorithms through rigorous proofs. In addition, we also study the unknown worker recruitment problem with fairness constraints. Here, the term “fairness” means that the platform must guarantee a minimum selection fraction for each registered worker, so that the platform can avoid the scenario where some workers are over-recruited but some others might be under-recruited. For this problem, we devise a fairness-aware unknown worker recruitment algorithm. Finally, we demonstrate the performance of the proposed algorithms through extensive simulations on real-world traces

The development of Blockchain-based mobile applications are impeded due to the resource limitations of mobile devices. Computation offloading can be a viable solution. In this paper, we consider a two-layer computation offloading paradigm including an edge computing service provider (ESP) and a cloud computing service provider (CSP). We formulate a multi-leader multi-follower Stackelberg game to address the computing resource management problem in such a network, by jointly maximizing the profits of each service provider (SP) and the payoffs of individual miners. We study two practical scenarios: a fixed-miner-number scenario for permissioned blockchains and a dynamic-miner-number scenario for permissionless blockchains. For the fixed-miner-number scenario, we discuss two different edge operation modes, i.e., the ESP isconnected(to the CSP) orstandalone, which form different miner subgames based on whether each miner's strategy set is mutually dependent. The existence and uniqueness of Stackelberg equilibrium (SE) in both modes are analyzed, according to which algorithms are proposed to achieve the corresponding SE(s). For the dynamic-miner-number scenario, we focus on the impact of population uncertainty and find that the uncertainty inflates the aggressiveness in the ESP resource purchasing. Numerical evaluations are presented to verify the proposed models.

Augmented reality (AR) applications that overlay the perception of the real world with digitally generated information are on the cusp of commercial viability. AR has appeared in several commercial platforms like Microsoft HoloLens and smartphones. They extend the user experience beyond two dimensions and supplement the normal 3D world of a user. A typical location-based multi-player AR application works through a three-step process, wherein the system collects sensory data from the real world, identifies objects based on their context, and finally, renders information on top of senses of a user. However, because these AR applications frequently exchange data with users, they have exposed new individual and public safety issues. In this paper, we develop ARSpy, a user location tracking system solely based on network traffic information of the user, and we test it on location-based multi-player AR applications. We demonstrate the effectiveness and efficiency of the proposed scheme via real-world experiments on 12 volunteers and show that we could obtain the geolocation of any target with high accuracy. We also propose three mitigation methods to mitigate these side channel attacks. Our results reveal a potential security threat in current location-based multi-player AR applications and serve as a critical security reminder to a vast number of AR users.

Voice-based input is usually used as the primary input method for augmented reality (AR) headsets due to immersive AR experience and good recognition performance. However, recent researches show that attackers can inject inaudible voice commands to the devices that lack voice verification. Even if we secure voice input with voice verification techniques, attackers can record the victim’s voice and replay it. To defend against voice-spoofing attacks, AR headsets should be able to determine whether the voice is from the person who is using the AR headsets. Existing voice-spoofing defense systems are designed for smartphone platforms and usually fail to work due to the special locations of microphones and loudspeakers on AR headsets. To address this challenge, in this paper, we propose a voice-spoofing defense system for AR headsets by leveraging both the internal body propagation and the air propagation of human voices. Experimental results show that our system can successfully accept normal users with average accuracy of 97 percent and defend against two basic types of attacks with average accuracy of at least 98 percent. More importantly, even if the attackers can fool our line-fitting model by manipulating special voice signals, our MCD-SVDD model can still reject them with accuracy of 100 percent.

Mobile CrowdSensing (MCS), through which a requester can coordinate a crowd of workers to accomplish some data collection tasks, has been recognized as a promising paradigm for large-scale data acquisition in recent years. Many researches focus on the worker recruitment problem in MCS, but most of them either have the assumption that workers’ qualities are known ahead of time or cannot ensure that workers report costs honestly. In this paper, we propose an incentive mechanism based on Combinatorial Multi-Armed Bandit and reverse Auction, called CMABA, to solve the multiple unknown workers recruitment problem in MCS. Our objective is to determine a recruiting strategy to maximize the total sensing quality under a limited budget, while ensuring truthfulness and individual rationality of sensing workers. We theoretically prove that our CMABA mechanism achieves truthfulness and individual rationality, and then analyze the regret of the mechanism. Based on CMABA, we ulteriorly propose an adaptive incentive mechanism, called ACMABA, to recruit workers via the alternative worker recruitment and quality update, which can achieve a higher total sensing quality and lower regret. Additionally, we also demonstrate significant performances of the CMABA and ACMABA mechanisms through extensive simulations on real-world data traces.

Cloud computing provides a feasible solution to data outsourcing, and hence forming a cloud-based data market, where data users buy data from owners through querying cloud servers. However, it also incurs new privacy and security problems, as data is under a centralized third-party instead of the data owner’s direct control. Existing data markets are also questioned on their inflexible and opaque pricing, where the value of data ownership and the cost of query searches are mixed. In this paper, we consider blockchain-based storage as a better choice to ensure safe data outsourcing since data is spread out across many data points. We propose an Ethereum-based data market that provides distributed storage and correct remote data search. We design a new pricing model, where each query will be charged by two parties: owner (paid for providing his data) and miner (rewarded by performing query searches). We study a new cooperative search scheme through a proxy to reduce cost on the user side. Given that each user query is charged based on its number of keywords, then a cooperative search can reduce user-side cost by combining multiple queries into a group so that overlapped keywords will only be charged for one time. To ensure user QoE, a combined query should not be significantly larger than any of its original queries in terms of the number of keywords. The total price is based on the total number of keywords in all groups. Since it is a cooperative model with shared resources, we also study various incentive properties on the user side, yielding a cost sharing mechanism to split joint cost in a truth-revealing and fair manner. We further extend our market with a set of substitute data owners and propose a double auction mechanism to match users and owners based on their requirements. Experiments have been conducted on real query trace to demonstrate the effectiveness of our proposed scheme.

Major cities worldwide have millions of cameras deployed for surveillance, business intelligence, traffic control, crime prevention, etc. Real-time analytics on video data demands intensive computation resources and high energy consumption. Traditional cloud-based video analytics relies on large centralized clusters to ingest video streams. With edge computing, we can offload compute-intensive analysis tasks to nearby servers, thus mitigating long latency incurred by data transmission via wide area networks. When offloading video frames from the front-end device to an edge server, the application configuration (i.e., frame sampling rate and frame resolution) will impact several metrics, such as energy consumption, analytics accuracy and user-perceived latency. In this paper, we study the configuration selection and bandwidth allocation for multiple video streams, which are connected to the same edge node sharing an upload link. We propose an efficient online algorithm, called JCAB, which jointly optimizes configuration adaption and bandwidth allocation to address a number of key challenges in edge-based video analytics systems, including edge capacity limitation, unknown network variation, intrusive dynamics of video contents. Our algorithm is developed based on Lyapunov optimization and Markov approximation, works online without requiring future information, and achieves a provable performance bound. We also extend the proposed algorithms to the multi-edge scenario in which each user or video stream has an additional choice about which edge server to connect. Extensive evaluation results show that the proposed solutions can effectively balance the analytics accuracy and energy consumption while keeping low system latency in a variety of settings.

With the integration of Artificial Intelligence (AI) and Internet of Things (IoT), the Federated Edge Learning (FEL), a promising computing framework is developing. However, there are still unsolved issues on communication efficiency and data security due to the huge models and unreliable transmission links. To address these issues, this paper proposes a novel federated edge learning system, called LightFed, where the edge nodes upload only vital partial local models, and successfully achieve lightweight communication and model aggregation. First, a novel model aggregation method Model Splitting and Splicing (MSS) and a Selective Parameter Transmission (SPT) scheme are proposed. By detecting the updating gradients of local parameters and filtering significant parameters, selective rotated transmission and efficient aggregation of local models are achieved. Second, a Training Filling Model (TFM) is proposed to infer the total data distribution of edge nodes, and train a filling model to mitigate the unbalanced training data without violating the data privacy of individual users. Moreover, a blockchain-powered confusion transmission mechanism is proposed for defending the attacks from external adversaries and protecting the model information. Finally, extensive experimental results demonstrate that our LightFed significantly outperforms the existing FEL systems in terms of communication efficiency and privacy security.

Mobile Edge Computing (MEC) has become an attractive solution to enhance the computing and storage capacity of mobile devices by leveraging available resources on edge nodes. In MEC, the arrivals of tasks are highly dynamic and are hard to predict precisely. It is of great importance yet very challenging to assign the tasks to edge nodes with guaranteed system performance. In this article, we aim to optimize the revenue earned by each edge node by optimally offloading tasks to the edge nodes. We formulate the revenue-driven online task offloading (ROTO) problem, which is proved to be NP-hard. We first relax ROTO to a linear fractional programming problem, for which we propose the Level Balanced Allocation (LBA) algorithm. We then show the performance guarantee of LBA through rigorous theoretical analysis, and present the LB-Rounding algorithm for ROTO using the primal-dual technique. The algorithm achieves an approximation ratio of$2(1+\xi)\ln (d+1)$with a considerable probability, where$d$is the maximum number of process slots of an edge node and$\xi$is a small constant. The performance of the proposed algorithm is validated through both trace-driven simulations and testbed experiments. Results show that our proposed scheme is more efficient compared to baseline algorithms.

The main method to achieve fault-tolerant network systems is by exploiting and effectively utilizing the edge-disjoint and/or inner-vertex-disjoint paths between pairs of source and destination vertices. Completely independent spanning trees (CISTs for short) are powerful tools for reliable broadcasting/unicasting and secure message distribution. Particularly, it has been shown that two CISTs have an application on configuring a protection routing in IP networks, such as mobile ad hoc networks and relatively large (static) network topologies with scalability in [IEEE/ACM Trans. Netw., 27 (2019) 1112-1123]. Many results focus on CISTs in specific networks in the literature, however, few results are given on an infinite class of networks having common properties. In this article, we prove the existence of dual-CISTs in an infinite number of networks satisfying some Hamilton sufficient conditions. A unique algorithm to construct a CIST-partition is proposed, which can be applied to not only many kinds of networks, but our algorithm can also be implemented very easily in parallel or distributed systems satisfying the conditions. In addition, we make a comparative analysis between the proposed conditions and several known results on an infinite number of networks, the advantage of our result is significant. In particular, the bound in our conditions is sharp. The results will provide a powerful framework for the design of fault-tolerant network topologies and routing protocols for future networks.

With the prevalence of cloud computing, a growing number of users are delegating clouds to host their sensitive data. To preserve user privacy, it is suggested that data is encrypted before outsourcing. However, data encryption makes keyword-based searches over ciphertexts extremely difficult. This is even challenging forfuzzy searchthat allows uncertainties or misspellings of keywords in a query. In this article, we propose a prime inner product encoding (PIPE) scheme, which makes use of theindecomposableproperty of prime numbers to provide efficient, highly accurate, and flexible multi-keyword fuzzy search. Our main idea is to encode either a query keyword or an index keyword into a vector filled with primes or reciprocals of primes, such that the result of vectors’ inner product is an integer only when two keywords are similar. Specifically, we first construct$\text{PIPE}_{0}$that is secure in the known ciphertext model. Unlike existing works that have difficulty supporting AND and OR semantics simultaneously,$\text{PIPE}_{0}$gives users the flexibility to specify different search semantics in their queries. Then, we construct$\text{PIPE}_{\text{S}}$that subtly adds random noises to a query vector to resist linear analyses. Both theoretical analyses and experiment results demonstrate the effectiveness of our scheme.

Cloud computing as a promising computing paradigm is increasingly utilized as potential hosts for users’ massive dataset. Since the cloud service provider (CSP) is outside the users’ trusted domain, existing research suggests encrypting sensitive data before outsourcing and adopting Searchable Symmetric Encryption (SSE) to facilitate keyword-based searches over the ciphertexts. However, it remains a challenging task to design an effective SSE scheme that simultaneously supportssublinear search time,efficient update and verification, andon-demand information retrieval. To address this, we propose a Verifiable Dynamic Encryption with Ranked Search (VDERS) scheme that allows a user to perform top-$K$Ksearches on adynamicdocument collection and verify the correctness of the search results in a secure and efficient way. Specifically, we first provide a basic construction,$\mathsf {VDERS}^0$VDERS0, where aranked inverted indexand averifiable matrixare constructed to enable verifiable document insertion in top-$K$Ksearches. Then, an advanced construction,$\mathsf {VDERS}^{\star }$VDERS★, is devised to further support document deletion with a reduced communication cost. Extensive experiments on real datasets demonstrate the efficiency and effectiveness of our VDERS scheme.

For distributed estimation arising in the nonlinear least squares (NLLSs) problems over adaptive networks, where every node has the abilities of data processing and learning, only the incomplete local data are exploited by the traditional noncooperative method, thereby resulting in the degradation on estimation performance. In this article, a cooperative diffusion strategy is proposed by using a Gauss–Newton (GN) method in order to fully utilize the diversity of temporal–spatial data on local updates. The proposed algorithm includes two steps, i.e., aggregate then update (ATU), where the aggregating step collects in real time the global information instead of local information due to the diffusion strategy, and the updating step implements the local GN iteration. The resulting ATU diffusion algorithm is a distributed and cooperative system without any increase on communication cost, as compared with the noncooperative version. Based on the detailed convergence analysis for ATU, which is fundamental to the promotion of this algorithm, the sufficient conditions for convergence are derived and the evidences of faster convergence than the noncooperative version are provided. The simulation results confirm the obtained theoretical derivations by applying the ATU algorithm to an NLLS-based target localization problem and show the cooperation gains in many aspects, such as the convergence rate, steady-state accuracy, and robustness to noisy range, step size, node, and link failures.

In dynamic spectrum access (DSA), secondary users (SU) should only be allowed to access a licensed band belonging to incumbent users (IU) when the quality-of-service (QoS) requirements of both IUs and SUs can be satisfied at the same time. However, IU’s location and its received interference strength are considered sensitive in many DSA systems which should not be revealed, making it very challenging to optimize the network utility subjected to satisfying the operation and security requirements of SUs and IUs. In this paper, we develop a secure and distributed SU transmit power control algorithm to solve this challenge. Our algorithm achieves optimal SU power control to maximize the sum of SU rates. The SINR-guaranteed coexistence between SUs and IUs are enabled to maintain effective communication, while no information is directly required from IUs. Local measurements of IU signals provided by Environmental sensing capability (ESC) also undergo a security masking process to ensure that IU location cannot be derived from its outputs. Convergence and stability properties of our algorithm and its privacy-protection strength are both theoretically analyzed and experimentally evaluated through simulations.

Mobile Edge Caching (MEC) can cache content at the edge of the network to reduce the delay and overhead of content transmission, which has become an effective method to solve the explosive growth of network traffic. To make good use of the limited resources in edge devices, many contents caching strategies use various methods to predict the popularity of content. However, caches get close to the edge of the network can lead to the rapid increase of caches' number and the user's requests are dispersed into a large number of caches, which leads to the popularity distribution of contents in edge caches is quite different and the number of first miss requests (the corresponding content is requested for the first time and is not in the cache) in edge caches becoming an essential factor affecting the cache hit rate. This paper first demonstrates the significant impact of the first miss requests through dataset analysis and establishes a mathematical model for the first miss problem in the edge cache. Then we analyze the similarity of requests received by caches and propose a proactive push algorithm based on similarity to improve the hit rate of edge caches. Through the trace-driven simulation experiment, we verify that the methods proposed in this paper can significantly improve the caches' hit rate.

Computation offloading is proposed to solve one obstacle of enabling high-accurate and real-time deep inference in resource-constrained Internet of Things (IoT) devices. Cooperative deep inference is proposed recently to further trade-off the introduced communication latency in computation offloading, which partitions a Deep Neural Network (DNN) model into two parts and utilizes the IoT end device and the server to process the DNN model cooperatively. We observe one important but ignored fact in all previous works: DNN computation and communication processing cbe conducted simultaneously in cooperative deep inference. As a result, the DNN layer-wise processing schedule has an impact on inference latency and it is non-trivial to find the optimal schedule in State-Of-The-Art (SOTA) DNNs with Directed Acyclic Graph (DAG) computational architectures. The contributions of this paper are as follows. (1) The proposed Deep Inference Optimization with Layer-wise Schedule, Deep-Inference-L, is a unique pipeline-based DAG schedule problem, which turns out to be NP-hard. (2) We categorize SOTA DNNs into three different categories and discuss the corresponding optimal processing schedule in special cases and efficient heuristic schedules in the general case. (3) The proposed solutions are extensively tested via a proof-of-concept prototype. (4) Results indicate that our algorithms can achieve an 8x speedup compared with local inference in the best case.

Cyber attacks are growing with the increase in internet usage. In a volumetric attack, the target resource is taken down with a huge amount of traffic. Distributed denial-of-service and link flooding attacks are examples of these types of attacks. In a non-volumetric attack, the attackers try to steal or get illegal authorization of some resources in a network. This type of attack can be severe even with a small amount of traffic. Non-volumetric attacks can be stopped by applying a moving target defense approach at the nodes on the attack path. An attack path is a series of steps and the attacker needs to succeed in all of those steps to gain access to the resources. In this paper, we propose an architecture to defend against both types of attacks. We formulate a problem to minimize the damage caused by the volumetric attack by using a limited number of blockage at some routers. This problem is NP-hard and we provide a greedy solution and provide an approximation ratio of it. We formulate another optimization problem to minimize the damage while securing the resources by deploying the minimum number of moving target defense methods. We provide a dynamic programming based solution to this problem. We conduct an extensive simulation to support our proposed models.

Reducing the inference time of Deep Neural Networks (DNNs) is critical when running time sensitive applications on mobile devices. Existing research has shown that partitioning a DNN and offloading a part of its computation to cloud servers can reduce the inference time. The single DNN partition problem has been extensively investigated recently. However, in real-world applications, a mobile device usually generates multiple DNN inference jobs simultaneously, and little attention has been paid to this case. We aim to minimize the makespan of multiple DNNs by jointly optimizing their partitioning and scheduling. Our observations show that the local computation time on a mobile device follows an increasing function, while the communication workload for offloading is usually decreasing as more DNN layers are computed. Based on this, we first relax our problem on continuous domain and show that partitioning all line-structure DNNs at the same layer is sufficient for makespan optimization. Then, for the discrete domain, two types of partitions are sufficient when the time difference between two adjacent partition layers is not drastic, subject to a given condition. An algorithm based on the binary search that efficiently finds optimal partition layers is illustrated. We also extend our approach to general-structure DNNs and offer a heuristic solution. Experiments have been conducted to evaluate the performance of different partition and scheduling methods on sample DNNs. Results validate the optimality of our theoretical results.

Vehicular CrowdSensing (VCS) has become a powerful sensing paradigm by selecting users driving vehicles to perform tasks. In most existing research, the platform centrally allocates tasks according to the collected user information. We argue that the information collection process results in user privacy leakage, and the centralized allocation leads to a heavy computation complexity. We propose to apply a distributed task allocation method in the widely-used route navigation system. The navigation system recommends several routes to a user and each route may cover some tasks. Then, the user distributively selects a route according to the route profit (task reward minus route cost). Since the task reward is shared by users, the route selections of users may influence each other. Hence, it remains unclear how to design a distributed route navigation approach to reach an equilibrium state (i.e., each user is satisfied with the selected route), while guaranteeing a good total profit. To this end, we formulate the problem as a multi-user potential game and propose a distributed route navigation algorithm. The trace-based simulation results verify that the proposed algorithm achieves a Nash equilibrium, while achieving a total user profit performance close to that of the optimal solution.

The multi-armed bandit (MAB) model has been deeply studied to solve many online learning problems, such as rate allocation in communication networks, Ad recommendation in social networks, etc. In an MAB model, given N arms whose rewards are unknown in advance, the player selects exactly one arm in each round, and his goal is to maximize the cumulative rewards over a fixed horizon. In this paper, we study the budget-constrained auction-based combinatorial multi-armed bandit mechanism with strategic arms, where the player can select K (<; N) arms in a round and pulling each arm has a unique cost. In addition, each arm might strategically report its cost in the auction. To this end, we combine the upper confidence bound (UCB) with auction to define the UCB-based rewards and then devise an auction-based UCB algorithm (called AUCB). In each round, AUCB selects the top K arms according to the ratios of UCB-based rewards to bids and further determines the critical payment for each arm. For AUCB, we derive an upper bound on regret and prove the truthfulness, individual rationality, and computational efficiency. Extensive simulations show that the rewards achieved by AUCB are at least 12.49% higher than those of state-of-the-art algorithms.

The quality of service (QoS) of intelligent applications on mobile devices heavily depends on the inference speed of Deep Neural Network (DNN) models. Cooperative DNN inference has become an efficient way to reduce inference latency. In cooperative inference, a mobile device offloads a part of its inference task to cloud servers. The large communication volume usually is the bottleneck of such systems. Priory research focuses on reducing the communication volume by finding optimal partition points. We notice that the computation and communication resources on mobile devices can work in pipeline, which can hide the communication time behind computation and further reduce the inference latency. Based on the observation, we formulate the offloading pipeline scheduling problem. We aim to find the optimal sequence of DNN execution and offloading for mobile devices such that the inference latency is minimized. If we use a directed acyclic graph (DAG) to model a DNN, the complex precedence constraints in DAGs bring challenges to our problem. Notice that most DNN models have independent paths or tree structures, we present an optimal path-wise DAG scheduler and an optimal layer-wise scheduler for tree-structure DAGs. Then, we proposed a heuristic based on topological sort to schedule general-structure DAGs. The prototype of our offloading scheme is implemented on a real-world testbed, where we use Raspberry Pi as the mobile device and lab PCs as the cloud. Various DNN models are tested and our scheme can reduce their inference latencies in different network environments.

Sparse spatiotemporal data completion is crucial in Mobile CrowdSensing for urban application scenarios. In fact, accurate urban data completion can enhance data expression, improve urban analysis, and ultimately guide city planning. However, it is a non-trivial task to consider outlier values caused by the special events (e.g., parking peak, traffic congestion, or festival parade) in spatiotemporal data completion because of the following challenges: 1) the rarity and unpredictability, 2) the inconsistency compared to normal values, and 3) the complex spatiotemporal relations. In spite of the considerable improvements, recent deep learning-based methods overlook the existence of outlier values, which results in misidentifying these values. To this end, focusing on spatiotemporal data, we propose a matrix completion method that takes outlier value effects into consideration. Specifically, an outlier value model is proposed by adding a memory network and modifying the loss function to traditional matrix completion. Along this line, we extract the features of outlier values and further efficiently complete and predict the unsensed data. Finally, we conduct both qualitative and quantitative experiments on three different datasets, and the results demonstrate that the performance of our method outperforms the state-of-the-art baselines.

Currently, the blockchain, as a key enabling technology of digital currency, has attracted lots of attention from both industry and academia. The blockchain mining process requires high computing power to solve a Proof-of-Work (PoW) puzzle, which is hard to implement on users’ mobile devices. So these miners may leverage the resources of the edge/cloud service providers (ESPs/CSP) to calculate the PoW puzzle. The existing edge-assisted blockchain networks simply assumed that all ESPs have a uniform propagation delay, which is not realistic. In this paper, we consider a more practical scene where ESPs with distributed geographic locations have diverse propagation delays when supporting the computation of the PoW puzzle. Additionally, the blockchain mining process generally involves the complicated competition and game among these ESPs and miners. Each ESP focuses on how to determine his resource price and to select the requests from the miners, so that he can maximize his utility. According to the set resource price, each miner concentrates on scheduling his resource requests for each ESP to maximize his individual utility which depends on ESPs’ resource price and propagation delays. We model such a resource pricing and scheduling problem as a multi-leader multi-follower Stackelberg game and aim at finding the joint maximization of the utilities of each ESP and each individual miner. We prove the existence and uniqueness of the Stackelberg equilibrium (SE) and meanwhile propose an algorithm to achieve the corresponding SE. Finally, extensive simulations are conducted to verify the significant performance of the proposed solution.

The emergence of federated learning and the increasingly powerful mobile devices lead to a mobile-crowd machine learning paradigm. In this paper, we consider a mobile-crowd federated learning system that includes a central server and a set of mobile devices. As the model requester, the server motivates all devices to train an accurate model by paying them based on their individual contributions. Each participating device needs to balance between the training rewards and costs for profit maximization. A Stackelberg game is proposed to model interactions between the server and devices. To match with reality, our model takes the training deadline and the device-side upload time into consideration. Based on different definitions of individual contribution, two reward policies, i.e., the size-based policy and accuracy-based policy, are compared. The existence and uniqueness of Stackelberg equilibrium (SE) under both definitions are analyzed, according to which algorithms are proposed to achieve the corresponding SE(s). We show that there is a lower bound of 0.5 on the price of anarchy in the proposed game. We extend our model by considering the uncertainty in the upload time, where each device’s upload time is subject to a normal distribution due to its unstable channel. Numerical evaluations are presented to verify the proposed models.

Bike-sharing systems (BSSs) are widely used in cities worldwide as they offer an affordable, eco-friendly method of transport. However, the rate of renting and returning bikes from stations is not always equal. The stations with imbalanced demand can become out of service by having all docks filled or emptied. These out-of-service stations can lead to a worse user experience and fewer people using BSSs. Researchers are trying to solve the rebalancing problem by developing algorithms that incentivize workers to pick up and drop off bicycles from different stations to balance the rent and return rates. Many of these algorithms focus on creating incentive and pricing models to encourage workers to go to imbalanced stations. Since they do not consider all of the placements of workers, this strategy may lead to inefficiencies where workers travel farther than they need. We can treat rebalancing as a Worker Assignment Problem by assigning worker stations to minimize the total distance traveled. We propose an algorithm that can approximate the optimal assignment significantly faster than other techniques with very high performance. The rapid speed allows for real-time use in augmenting pricing models and as a stand-alone method for worker assignment. Furthermore, we compare our approach against four existing algorithms on real-world data to evaluate computational speed and effectiveness.

Bike-Sharing Systems (BSS) is a new mode of transportation that allows users to rent bicycles and use them across the city for short trips. It has spread to all of the continents in the world due to its convenient nature and benefits. In contrast to other methods of public transportation, this service allows users to travel directly to their desired location, solving the first/last mile problem—when users’ public transportation does not take them within one-fourth of a mile from their final destination. This allows users to get closer to their final destination than taking the subway alone, encouraging the use of public transportation. BSS also increases physical exercise, reduces carbon emissions, and is a flexible mode of transportation. These benefits have led to the growth of BSS. With this growth, several problems have emerged. These include re-balancing of underflow and overflow stations, predicting demand for each station, station placement, and, most importantly, safety. Re-balancing, predicting station demand, and station placement have been widely studied; however, safety has not been widely studied. This paper discusses the four leading safety concerns and provides new possible solutions and improvements to current solutions.

One of the main hindrances that wireless sensor networks (WSNs) face is the battery-powered sensors that need to be charged from time to time. Recently, the approach of having mobile chargers (MCs) that travel to the static sensors of the network and transfer energy wirelessly and efficiently has become a promising solution to that hindrance. An optimization problem, called the mobile charger coverage problem, arises naturally to keep all of the sensors alive with an objective of determining both the minimum number of MCs required to meet the sensor recharge frequency and the schedule of these MCs. It is shown that this optimization problem becomes NP-hard in a high-dimensional space, including 2-D space with homogeneous recharging frequency. On the other hand, it is shown that a polynomial-time algorithm exists for sensors with a homogeneous recharge frequency on a 1-dimensional space (line or ring). In this paper, we seek to find a delicate border between the tractable and intractable problem space. Specifically, we study the special case of heterogeneous sensors that take frequencies of 1’s and 2’s (lifetimes of 1 and 0.5 time units) on a line, conjecture its NP-hardness, propose a novel brute-force optimal algorithm, and present a linear-time greedy algorithm that gives a 1.5-approximation solution for the problem. A comprehensive simulation is conducted to verify the efficiency of using our proposed algorithms.

Link congestion due to regular traffic and link flooding attacks (LFA) are two major problems in datacenters. Recent usage growth of software defined networking (SDN) in datacenters enables dynamic and convenient configuration management that makes it easy to reconfigure the network to mitigate the LFA. The reconfiguration that redirects some of the traffic can be done in two ways: the shortest alternative path and the minimum changes in rule path. The SDN switches have a limited capacity for the rules and the performance dramatically drops when the number of stored rules is higher. Besides, it takes some time to adopt the changes by the SDN switches which causes interruption in flow. In this paper, we aim at minimizing the number of rule changes while redirecting some of the traffic from the congested link. We formulate two problems to minimize the number of rule changes to redirect traffic. The first problem is the basic and it considers a congested link and a flow to direct. We provide a Dijkstra-based and a rule merging based solution to the problems. The second problem considers multiple flows and we propose flow grouping and rule merging based solutions. We conduct extensive simulations and experiments in our datacenter to support our model.

Dynamic searchable symmetric encryption (DSSE) that enables a client to perform searches and updates on encrypted data has been intensively studied in cloud computing. Recently, forward privacy and backward privacy has engaged significant attention to protect DSSE from the leakage of updates. However, the research in this field almost focused on keyword-level updates. That is, the client needs to know the keywords of the documents in advance. In this paper, we proposed a document-level update scheme, DBP, which supports immediate deletion while guaranteeing forward privacy and backward privacy. Compared with existing forward and backward private DSSE schemes, our DBP scheme has the following merits: 1) Practicality. It achieves deletion based on document identifiers rather than document/keyword pairs; 2) Efficiency. It utilizes only lightweight primitives to realize backward privacy while supporting immediate deletion. Experimental evaluation on two real datasets demonstrates the practical efficiency of our scheme.

Summary Fair allocation has been studied intensively in both economics and computer science. Many existing mechanisms that consider fairness of resource allocation focus on a single resource. With the advance of cloud computing that centralizes multiple types of resources under one shared platform, multi‐resource allocation has come into the spotlight. In fact, fair/efficient multi‐resource allocation has become a fundamental problem in any shared computer system. The widely used solution is to partition resources into bundles that contain fixed amounts of different resources, so that multiple resources are abstracted as a single resource. However, this abstraction cannot satisfy different demands from heterogeneous users, especially on ensuring fairness among users competing for resources with different capacity limits. A promising approach to this problem is dominant resource fairness (DRF), which tries to equalize each user's dominant share (share of a user's most highly demanded resource, that is, the largest fraction of any resource that the user has required for a task), but this method may still suffer from significant loss of efficiency (i.e., some resources are underused). This article develops a new allocation mechanism based on DRF aiming to balance fairness and efficiency. We consider fairness not only in terms of a user's dominant resource, but also in another resource dimension which is secondarily desired by this user. We call this allocation mechanism 2‐dominant resource fairness (2‐DF). Then, we design a non‐trivial on‐line algorithm to find a 2‐DF allocation and extend this concept to k‐dominant resource fairness (k‐DF).

To achieve security in wireless sensor networks (WSNs), it is important to be able to encrypt messages sent among sensor nodes by using shared keys between them. Due to resource constraints, achieving such key agreement in WSNs is nontrivial. Previous research indicates that key management schemes using deployment knowledge can significantly improve the performance of WSNs. Nevertheless, in these schemes, resilient local connectivity and resilient global connectivity become unstable when deployment error changes. To resolve the above problem, in this article, a one-way associated key management model is proposed. In this model, the key pool consists of two layers: 1) the global layer and 2) the local layer. According to different deployment errors, the number of keys allocated from the global key pool and local key pools can be dynamically adjusted, thereby improving the stability of networks' performance. In multiphase sensor networks, analysis and simulation indicate that our scheme has better adaptability in applications where deployment error changes as compared with related schemes.

Recent research advancement of wireless sensing technology has made device-free interaction in the WiFi-enabled IoT environment possible. Although gesture-based interaction with such a smart environment greatly improves usability, it also introduces many security problems, such as shoulder surfing attacks. By spoofing the gestures of legitimate users, the attacker could easily access private information or services and cause even worse consequences. A secure interaction mechanism for this environment is required to prevent attackers without compromising the usability, while the limited recognition ability and low robustness of WiFi sensing make this target extremely challenging. To this end, we propose a secure interaction mechanism called secure interaction via WiFi Signal (SiWi), which provides the ability to resist shoulder surfing attacks without compromising the usability by using just WiFi signals. SiWi innovates in a concurrent interaction/authentication framework with only three elemental gestures (push, swing, and wave) and four types of identity-related imperceptible/hidden features (time distribution, direction, angle, and distance). HMM and Fresnel model-based algorithms are used to recognize the gestures and extract hidden features robustly and efficiently. Extensive experiments in a real implemented system were conducted to investigate the effectiveness of the proposed secure interaction system. The results show that our system can achieve an average accuracy of 93% to identify legitimate users and 97% to resist the spoofer.

UAV-assisted wireless communications facilitate the applications of Internet of Things (IoT), which employ billions of devices to sense and collect data with an on-demand style. However, there are numerous malicious Mobile Data Collectors (MDCs) mixing into the network, stealing or tampering with data, which greatly damages IoT applications. So, it is urgent to build a ubiquitous trust communication system. In this paper, a UAV-assisted Ubiquitous Trust Evaluation (UUTE) framework is proposed, which combines the UAV-assisted global trust evaluation and the historical interaction based local trust evaluation. We first propose a global trust evaluation model for data collection platforms. It can accurately eliminate malicious MDCs and create a clean data collection environment, by dispatching UAVs to collect baseline data to validate the data submitted by MDCs. After that, a local trust evaluation model is proposed to help select credible MDCs for collaborative data collection. By letting UAVs distribute the data verification hash codes to MDCs, the MDCs can verify whether the exchanged data from the interacted MDCs is reliable. Extensive experiments conduct on a real-life dataset demonstrate that our UUTE system outperforms the existing trust evaluation systems in terms of accuracy and cost.

Offloading cellular traffic via Device-to-Device communication (or D2D offloading) has been proved to be an effective way to ease the traffic burden of cellular networks. However, mobile nodes may not be willing to take part in D2D offloading without proper financial incentives since the data offloading process will incur a lot of resource consumption. Therefore, it is imminent to exploit effective incentive mechanisms to motivate nodes to participate in D2D offloading. Furthermore, the design of the content caching strategy is also crucial to the performance of D2D offloading. In this paper, considering these issues, a novel Incentive-driven and Deep Q Network (DQN) based Method, named IDQNM is proposed, in which the reverse auction is employed as the incentive mechanism. Then, the incentive-driven D2D offloading and content caching process is modeled as Integer Non-Linear Programming (INLP), aiming to maximize the saving cost of the Content Service Provider (CSP). To solve the optimization problem, the content caching method based on a Deep Reinforcement Learning (DRL) algorithm, named DQN is proposed to get the approximate optimal solution, and a standard Vickrey-Clarke-Groves (VCG)-based payment rule is proposed to compensate for mobile nodes' cost. Extensive real trace-driven simulation results demonstrate that the proposed IDQNM greatly outperforms other baseline methods in terms of the CSP's saving cost and the offloading rate in different scenarios.

With the progress of mobile devices and the successful forms of using the wisdom of crowds, spatial crowdsourcing has attracted much attention from the research community. The idea of spatial crowdsourcing is recruiting a set of available crowds to finish the spatial tasks located in crowdsourcing locations, e.g., landmarks, by using their handheld devices. This paper addresses the worker recruitment problem in spatial crowdsourcing under the coverage and workload-balancing requirements. The coverage constraint means that any crowdsourcing location should be visited by at least one of the recruited workers to satisfy the Quality-of-Service requirement, e.g., traffic monitoring or climate forecast. In addition, we argue that each crowdsourcing operation has a cost in reality, e.g., data traffic or energy consumption and the resource may be limited at each crowdsourcing location. The objective of this paper is to solve a Coverage and Balanced Crowdsourcing Recruiting (CBCR) problem, which ensures the coverage requirement and minimizes the maximum crowdsourcing cost for any crowdsourcing location. We prove that the CBCR problem is NP-hard in the general case. Then, we discuss the CBCR problem in the 1-D scenario. In the 1-D scenario, we first propose a directionally coverage scheme and further extend it to a Polynomial-Time Approximation Scheme (PTAS) to trade-off the computation complexity and the performance. The performance can be bounded to 2 + ε, where ε can be an arbitrary small value. Then, we found that there exists a sub-optimal structure, and thus the dynamic programming approach is proposed to find the optimal solution in the 1-D scenario. In the general 2-D scenario, we first prove that it has a sub-modular property and thus the naive greedy algorithm has an approximation ratio of lnn+ 1. In addition, we propose a randomized rounding algorithm with an expectation bound of O(logn/ log logn). Extensive experiments on realistic traces demonstrate the effectiveness of the proposed algorithms.

Monitoring all the internal flows in a datacenter is important to protect a victim against internal distributed denial-of-service (DDoS) attacks. Unused virtual machines (VMs) in a datacenter are used as monitors and flows are copied to the monitors from software defined networking (SDN) switches by adding some special rules. In such a system, a VM runs a machine learning method to detect DDoS behavior but it can only process a limited number/amount of flows. When the amount of flows is beyond the capacities of all monitor VMs, the system sub-samples each flow probabilistically. The sampling rate affects the DDoS detection rate of the monitors. Besides, the DDoS detection rates of different types of flows are different for the same sampling rate. A uniform sampling rate might not produce a good overall DDoS detection rate. Assigning different sampling rates to different flows may produce the best result. In this paper, we propose a flow grouping approach based on behavioral similarity among the VMs followed by hierarchical clustering of VMs. The sampling rate is uniform among all the flows in a group. We investigate the relationship between the sampling rate and the DDoS detection rate. Then, we formulate an optimization problem for finding an optimal sampling rate distribution and solve it using mix-integer linear programming. We conduct extensive experiments with Hadoop and Spark and present results that support the feasibility of our model.

The gap between the peak-hour Internet and the average level is increasing, which inevitably creates a type of temporary cellular weak coverage when there is a surge in data traffic demand, where any cell-edge device will have a low spectrum efficiency (SE). In this article, we propose a novel incentive architecture based on the dynamic radio frequency charging technology to improve the SE and use the Stackelberg game theory to formulate the problem. In such a model, a small base station (SBS) acts as the leader to offer a desired partition of the resource block obtained by a cell-edge device, while some small energy providers (SEPs) and small virtual access points (SVAPs) that are selected from user equipment act as the followers to make their decisions, respectively, to compete for the free part of such a resource block. Following the potential game rules, all the SEPs compete for a specific free resource part allocated by the SBS, and then, all the SVAPs compete for another nonoverlapping part allocated by the SBS on the basis of the results of the SEPs' potential game. Although our incentive architecture formally has three game stages, it is essentially a two-level Stackelberg game, which is analyzed by using a backward induction method. The theoretical analysis proves the convergence of the above-mentioned game models, and the simulation results demonstrate that the proposed incentive architecture can improve the SE for each cell-edge device.

Online reviews and ratings play an important role in shaping the purchase decisions of customers in e-commerce. Many researches have been done to make proper recommendations for users, by exploiting reviews, ratings, user profiles, or behaviors. However, the dynamic evolution of user preferences and item properties haven’t been fully exploited. Moreover, it lacks fine-grained studies at the aspect level. To address the above issues, we define two concepts of user maturity and item popularity, to better explore the dynamic changes for users and items. We strive to exploit fine-grained information at the aspect level and the evolution of users and items, for dynamic sentiment prediction. First, we analyze three real datasets from both the overall level and the aspect level, to discover the dynamic changes (i.e., gradual changes and sudden changes) in user aspect preferences and item aspect properties. Next, we propose a novel model of Aspect-based Sentiment Dynamic Prediction (ASDP), to dynamically capture and exploit the change patterns with uniform time intervals. We further propose the improved model ASDP+ with a bin segmentation algorithm to set the time intervals non-uniformly based on the sudden changes. Experimental results on three real-world datasets show that our work leads to significant improvements.

Distributed Interactive Applications (DIAs) such as virtual reality and multiplayer online game usually require fast processing of tremendous data and timely exchange of delay-sensitive action data and metadata. This makes traditional mobile-based or cloud-based solutions no longer effective. Thanks to edge computing, DIA Service Providers (DSPs) can rent resources from Edge Infrastructure Providers (EIPs) to place service entities that store user states and run computation-intensive tasks. One fundamental problem for a DSP is to decide where to place service entities to achieve low-delay pairwise interactions between DIA users, under the constraint that the total placement cost is no more than a specified budget threshold. In this article, we formally model the service entity placement problem and prove that it is NP-complete by a polynomial reduction from the set cover problem. We present GPA, an efficient algorithm for service entity placement, and theoretically analyze its performance. We evaluated GPA with both real-world data trace-driven simulations, and observed that GPA performs close to the optimal algorithm and generally outperforms the baseline algorithm. We also output a curve showing the trade-off between the weighted average interaction delay and the budget threshold, so that a DSP can choose the right balance.

Mobile crowdsensing is a new paradigm by which a platform can recruit mobile workers to perform some sensing tasks by using their smart mobile devices. In this paper, we focus on a privacy-preserving unknown worker recruitment issue. The platform needs to recruit some workers without knowing the qualities of them completing tasks. Meanwhile, these quality information also needs to be protected from disclosure. To tackle these challenges, we model the unknown worker recruitment as a Differentially Private Multi-Armed Bandit (DP-MAB) game by seeing each worker as an arm of DP-MAB and the task completion quality contributed by each worker as the reward of pulling arm. Then, recruiting workers is equivalent to designing a bandit policy of pulling DP-MAB arms. Under this model, we propose a Differentially Private ϵ-First-based arm-pulling (DPF) algorithm and a Differentially Private UCB-based arm-pulling (DPU) algorithm, which can achieve the nearly optimal expected accumulative rewards under a given budget. We also analyze the regrets of the DPF and DPU algorithms and prove that both of them are δ-differentially private on the task completion qualities (δ > 0δ). Finally, we conduct extensive simulations to verify the significant performances of DPF and DPU based on both the real-trace and synthetic datasets.

With the ever-increasing amount of data resided in a cloud, how to provide users with secure and practical query services has become the key to improve the quality of cloud services. Fuzzy searchable encryption (FSE) is identified as one of the most promising approaches for enabling secure query services, since it allows searching encrypted data by using keywords with spelling errors. However, existing FSE schemes are far from the practical use for the following reasons: (1)Inflexibility.It is hard for them to simultaneously support AND and OR semantics in a multi-keyword query. (2)Inefficiency.They require sequentially scanning a whole dataset to find matched files, and thus are difficult to apply to a large-scale dataset. (3)Limited robustness.It is difficult for them to resist the linear analysis attack in the known-background model. To fix the above problems, this article proposes matrix-based multi-keyword fuzzy search (M2FS) schemes, which support approximate keyword matching by exploiting the indecomposable property of primes. Specifically, we first present a basic scheme, called M2FS-B, where multiple keywords in a query or a file are constructed as prime-related matrices such that the result of matrix multiplication can be employed to determine the level of matching for different query semantics. Then, we construct an advanced scheme, named M2FS-E, which builds a searchable index as a keyword balanced binary (KBB) tree for dynamic and parallel searches, while adding random noises into a query matrix for enhanced robustness. Extensive analyses and experiments demonstrate the validity of our M2FS schemes.

Mobile crowdsensing has become a significant sensing technique which takes advantage of mobile devices to collect information about the surrounding. The traditional cloud-based centralized mobile crowdsensing architecture generates significant traffic on networks and computation burden on the cloud. In this paper, we investigate the edge-based mobile crowdsensing architecture, where a group of mobile edge servers is deployed at network edge as the bridge between the central server and mobile users for data filtering and aggregation. Each user may collect multiple types of data in mobile crowdsensing. To facilitate data aggregation, the same type of data carried by different users is supposed to be uploaded to the same mobile edge server. In this scenario, a problem emerges: which server should be activated for processing each type of data in order to minimize the total cost? The cost consists of the facility cost (activating server and processing data) and the service cost (the users' movement cost for uploading data). Furthermore, the problem is formulated as a variant of the uncapacitated multi-commodity facility location problem. In particular, two situations of the problem are studied in our work: (1) for the situation where each user carries at most two types of data, we propose a relaxation based approximation algorithm, which is proved to have a bound to the optimal solution; (2) for a more generalized situation where each user can carry multiple types of data, we propose a connected multi-agent simulated annealing algorithm. Finally, we conduct extensive simulations based on the widely-used real-world datasets: roma/taxi, epfl/mobility and geolife trajectory. The simulation results show that the proposed algorithms demonstrate their superiority over baseline methods and are consistent with the theoretical analysis.

Mobile CrowdSensing (MCS) has recently become a powerful paradigm that recruits users to cooperatively perform various tasks. In many realistic settings, users participate in real time and we have to recruit them in an online manner. The existing works usually formulate the online recruitment problem as a budgeted optimal stopping problem with submodular user utility, while we first argue that not only the budget but also the time constraints can jointly influence the recruitment performance. For example, if we have less budget but plenty of time, we should recruit users with more patience. Second, considering the user’s cooperative willingness, its contribution may be diminishing or even irregular. Hence, we also need to address not only submodular cases but also their non-submodular utility. In this paper, we study the online user recruitment problem with (non-)submodular utility under the budget and time constraints. To deal with the two constraints, we first estimate the number of users to be recruited and then recruit them in segments. Moreover, we extend the segmented strategy with a non-submodular utility, which has the submodularity ratio$\gamma $and the competitive ratio$\gamma ^{2}(1-e^{-1})/7$. Furthermore, to correct estimation errors and utilize newly obtained information, we dynamically re-adjust the segmented strategy and also prove that the dynamic strategy achieves a competitive ratio of${\gamma ^{2}(1-e^{-1})(1-e^{-\gamma /2})/7}$. Finally, a reverse auction-based online pricing mechanism is lightly built into the proposed user recruitment strategy, which achieves truthfulness and individual rationality. Extensive experiments on three real-world data sets validate the proposed online user recruitment strategy under the (non-) submodular utility and two constraints.

Instead of relying on remote clouds, today's Augmented Reality (AR) applications usually send videos to nearby edge servers for analysis (such as objection detection) so as to optimize the user's quality of experience (QoE), which is often determined by not only detection latency but also detection accuracy, playback fluency, etc. Therefore, many studies have been conducted to help adaptively choose best video configuration, e.g., resolution and frame per second (fps), based on network bandwidth to further improve QoE. However, we notice that the video content itself has significant impacts on the configuration selection, e.g., the videos with high-speed objects must be encoded with a high fps to meet the user's fluency requirement. In this article, we aim to adaptively select configurations that match the time-varying network condition as well as the video content. We design Cuttlefish, a system that generates video configuration decisions using reinforcement learning (RL). Cuttlefish trains a neural network model that picks a configuration for the next encoding slot based on observations collected by AR devices. Cuttlefish does not rely on any pre-programmed models or specific assumptions on the environments. Instead, it learns to make configuration decisions solely through observations of the resulting performance of historical decisions. Cuttlefish automatically learns the adaptive configuration policy for diverse AR video streams and obtains a gratifying QoE. We compared Cuttlefish to several state-of-the-art bandwidth-based and velocity-based methods using trace-driven and real world experiments. The results show that Cuttlefish achieves a 18.4-25.8 percent higher QoE than the others.

The booming of Convolutional Neural Networks (CNNs) has empowered lots of computer-vision applications. Due to its stringent requirement for computing resources, substantial research has been conducted on how to optimize its deployment and execution on resource-constrained devices. However, previous works have several weaknesses, including limited support for various CNN structures, fixed scheduling strategies, overlapped computations, high synchronization overheads, etc. In this article, we present DeepSlicing, a collaborative and adaptive inference system that adapts to various CNNs and supports customized flexible fine-grained scheduling. As a built-in functionality, DeepSlicing has supported typical CNNs including GoogLeNet, ResNet, etc. By partitioning both model and data, we also design an efficient scheduler, Proportional Synchronized Scheduler (PSS), which achieves the trade-off between computation and synchronization. Based on PyTorch, we have implemented DeepSlicing on the testbed with real-world edge settings that consists of 8 heterogeneous Raspberry Pi's. The results indicate that DeepSlicing with PSS outperforms the existing systems dramatically, e.g., the inference latency and memory footprint are reduced up to 5.79× and 14.72×, respectively.

Edge cloud computing is a new paradigm in which the computation and storage services of remote cloud data centers are moved to Edge Cloud Nodes (ECNs) in network edges. Compared to traditional cloud data centers, ECNs are geographically close to mobile users so the communication latency is significantly reduced. In this paper, we study the problem of allocating Virtual Machine (VM) resources in geo-distributed ECNs to mobile users by using the auction theory. First, we treat mobile users and ECNs as the buyers and sellers of the VM resource auction, respectively. Then, we model the VM resource allocation problem as an$n$-to-one weighted bipartite graph matching problem with 0-1 knapsack constraints. Since this problem is NP-hard, we design a greedy approximation algorithm to determine the winners of the auction, based on which we propose a truthful Auction-based VM resource Allocation (AVA) mechanism to solve the problem. Moreover, we prove that the AVA mechanism not only achieves an approximately optimal solution for winner selection, but also has the properties of truthfulness, individual rationality, and computational efficiency. Finally, we conduct extensive simulations on real traces to verify the significant performances of the proposed AVA mechanism.

A large number of servers are interconnected using a specific datacenter network to deliver the infrastructure as a service (IaaS). Multicast can jointly utilize the network resources and further reduce the consumption of network bandwidth more than individual unicast. The source of a multicast service, however, does not need to be in a specific location as long as certain constraints are satisfied. This means the multicast can have uncertain sources, which could reduce the network resource consumption more than a traditional multicast service and further improve the quality of service. In this paper, we propose a novel reliable multicast service with uncertain sources named ReMUS. The goal is to minimize the sum of the transfer cost and the recovery cost, although finding such a ReMUS is very challenging. Thus, we design a source-based multicast method to solve this problem by exploiting the flexibility of sources when no recovery nodes exist in the network. Furthermore, we design a general multicast method to jointly exploit the benefits of uncertain sources and recovery nodes to minimize the total cost of ReMUS. We conduct extensive evaluations under Internet2 and datacenter networks. The results indicate that our methods can efficiently realize the reliable and scalable multicast with uncertain sources, irrespective of the settings of networks and multicasts. To the best of our knowledge, we are the first to study the reliable multicast service under uncertain sources.

The e-healthcare cloud system has shown its potential to improve the quality of healthcare and individuals' quality of life. Unfortunately, security and privacy impede its widespread deployment and application. There are several research works focusing on preserving the privacy of the electronic healthcare record (EHR) data. However, these works have two main limitations. First, they only support the `black or white' access control policy. Second, they suffer from the inference attack. In this paper, for the first time, we design an inference attack-resistant e-healthcare cloud system with fine-grained access control. We first propose a two-layer encryption scheme. To ensure an efficient and fine-grained access control over the EHR data, we design the first-layer encryption, where we devise a specialized access policy for each data attribute in the EHR, and encrypt them individually with high efficiency. To preserve the privacy of role attributes and access policies used in the first-layer encryption, we systematically construct the second-layer encryption. To take full advantage of the cloud server, we propose to let the cloud execute computationally intensive works on behalf of the data user without knowing any sensitive information. To preserve the access pattern of data attributes in the EHR, we further construct a blind data retrieving protocol. We also demonstrate that our scheme can be easily extended to support search functionality. Finally, we conduct extensive security analyses and performance evaluations, which confirm the efficacy and efficiency of our schemes.

MapReduce includes three phases of map, shuffle, and reduce. Since the map phase is CPU-intensive and the shuffle phase is I/O-intensive, these phases can be conducted in parallel. This paper studies a joint scheduling optimization of overlapping map and shuffle phases to minimize the average job makespan. New concepts of the strong pair and the weak pair are introduced. Two jobs are defined as a strong pair if the shuffle and map workloads of one job equal the map and shuffle workloads of the other job, respectively. Two jobs are defined as a weak pair if their total map workloads equal their total shuffle workloads. We prove that if the entire set of jobs can be decomposed to strong pairs of jobs, then the optimal schedule can pairwisely execute jobs that can form a strong pair. Following the above intuition, several offline and online scheduling policies are proposed. Extensions are made based on weak pairs. Real data-driven experiments validate the efficiency and effectiveness of the proposed policies.

In online systems, including e-commerce platforms, many users resort to the reviews or comments generated by previous consumers for decision making, while their time is limited to deal with many reviews. Therefore, a review summary, which contains all important features in user-generated reviews, is expected. In this article, we study “how to generate a comprehensive review summary from a large number of user-generated reviews.” This can be implemented by text summarization, which mainly has two types of extractive and abstractive approaches. Both of these approaches can deal with both supervised and unsupervised scenarios, but the former may generate redundant and incoherent summaries, while the latter can avoid redundancy but usually can only deal with short sequences. Moreover, both approaches may neglect the sentiment information. To address the above issues, we propose comprehensive Review Summary Generation frameworks to deal with the supervised and unsupervised scenarios. We design two different preprocess models of re-ranking and selecting to identify the important sentences while keeping users’ sentiment in the original reviews. These sentences can be further used to generate review summaries with text summarization methods. Experimental results in seven real-world datasets (Idebate, Rotten Tomatoes Amazon, Yelp, and three unlabelled product review datasets in Amazon) demonstrate that our work performs well in review summary generation. Moreover, the re-ranking and selecting models show different characteristics.

Event-based social networks (EBSNs) connect online and offline lives. They allow online users with similar interests to get together in real life. Attendance prediction for activities in EBSNs has attracted a lot of attention and several factors have been studied. However, the prediction accuracy is not very good for some special activities, such as outdoor activities. Moreover, a very important factor, the weather, has not been well exploited. In this work, we strive to understand how the weather factor impacts activity attendance, and we explore it to improve attendance prediction from the organizer’s view. First, we classify activities into two categories: the outdoor and the indoor activities. We study the different ways that weather factors may impact these two kinds of activities. We also introduce a new factor of event duration. By integrating the above factors with user interest and user-event distance, we build a model of attendance prediction with the weather named GBT-W , based on the Gradient Boosting Tree. Furthermore, we develop a platform to help event organizers estimate the possible number of activity attendance with different settings (e.g., different weather, location) to effectively plan their events. We conduct extensive experiments, and the results show that our method has a better prediction performance on both the outdoor and the indoor activities, which validates the reasonability of considering weather and duration.

With the rapid popularity of cameras on various devices, video chat has become one of the major ways for communication, such as online meetings. However, the recent progress of face reenactment techniques enables attackers to generate fake facial videos and use others' identities. To protect video chats against fake facial videos, we propose a new defense system to significantly raise the bar for face reenactment-assisted attacks. Compared with existing works, our system has three major strengths. First, our system does not require extra hardware or intense computational resources. Second, it follows the normal video chat process and does not significantly degrade the user experience. Third, our system does not need to collect training data from attackers and new users, which means it can be quickly launched on new devices. We developed a prototype and conducted comprehensive evaluations. Experimental results show that our system can provide an average true acceptance rate of at least 92.5% for legitimate users and reject the attacker with mean accuracy of at least 94.4% for a single detection.

In this paper, we consider a fundamental problem: given one mobile charger that can charge multiple sensor nodes simultaneously, how we can schedule it to charge a given WSN to maximize the energy usage effectiveness (EUE)? We propose a novel charging paradigm-Overlapped Mobile Charging (OMC)- the first of its kind to the best of our knowledge. Firstly, OMC clusters sensor nodes into multiple non-overlapped sets using k-means evaluated by the Davies-Bouldin Index, such that the sensor nodes in each set have similar recharging cycles. Secondly, for each set of sensor nodes, OMC further divides them into multiple overlapped groups, and charges each group at different locations for different time durations to make sure that each overlapped sensor node just receives its required energy from multiple charging locations.

With the development of deep learning, artificial intelligence applications and services have boomed in the recent years, including recommendation systems, personal assistant and video analytics. Similar to other services in the edge computing environment, artificial intelligence computing tasks are pushed to the network edge. In this paper, we consider the multi-user edge-assisted video analytics task offloading (MEVAO) problem, where users have video analytics tasks with various accuracy requirements. All users independently choose their accuracy decisions, satisfying the accuracy requirement, and offload the video data to the edge server. With the utility function designed based on the features of video analytics, we model MEVAO as a game theory problem and achieve the Nash equilibrium. For the flexibility of making accuracy decisions under different circumstances, a deep reinforcement learning approach is applied to our problem. Our proposed design has much better performance compared with some other approaches in the extensive simulations.

Mobile CrowdSensing (MCS) is a promising paradigm that recruits mobile users to cooperatively perform various sensing tasks. When assigning tasks to users, most existing works only consider the fairness of users, i.e., the user's processing ability, with the goal of minimizing the assignment cost. However, in this paper, we argue that it is necessary to not only give full use of all the users' ability to process the tasks (e.g., not exceeding the maximum capacity of each user while also not letting any user idle too long), but also satisfy the assignment frequency of all corresponding tasks (e.g., how many times each task should be assigned within the whole system time) to ensure a long-term, double-fair and stable participatory sensing system. Hence, to solve the task assignment problem which aims to reasonably assign tasks to users with limited task processing ability while ensuring the assignment frequency, we first model the two fairness constraints simultaneously by converting them to user processing queues and task virtual queues, respectively. Then, we propose a Fair Task Assignment Strategy (FTAS) utilizing Lyapunov optimization and we provide the proof of the optimality for the proposed assignment strategy to ensure that there is an upper bound to the total assignment cost and queue backlog. Finally, extensive simulations have been conducted over three real-life mobility traces: Changchun/taxi, Epfl/mobility, and Feeder. The simulation results prove that the proposed strategy can achieve a trade-off between the objective of minimizing the cost and the fairness of tasks and users compared with other baseline approaches.

The implementation of network services is changed from dedicated hardware to software middleboxes with the evolution of Network Function Virtualization (NFV). The placement of such middleboxes are complicated not only by the selection of multiple available hosting servers, but also by the traffic-changing effect of middleboxes. In this paper, we address the placement problem of a single type of traffic-diminishing middlebox (e.g., spam filters), where the objective is to minimize the total bandwidth consumption when the total number of placed middleboxes is limited. We prove the NP-hardness of checking the feasibility of our problem in general topologies. Then we propose a greedy solution and prove that it is performance-guaranteed when it generates a feasible deployment. Next we narrow down to tree-structured networks and propose an optimal dynamic programming based strategy. In order to improve the time efficiency, we also introduce an efficient greedy solution with an intuitive insight. Extensive simulations are conducted on a real-world dataset to evaluate the performance of our algorithms.

Mobile crowdsensing, through which a requester can coordinate a crowd of workers to complete some sensing tasks, has attracted significant attention recently. In this paper, we focus on the unknown worker recruitment problem in mobile crowdsensing, where workers' sensing qualities are unknown a priori. We consider the scenario of recruiting workers to complete some continuous sensing tasks. The whole process is divided into multiple rounds. In each round, every task may be covered by more than one recruited workers, but its completion quality only depends on these workers' maximum sensing quality. Each recruited worker will incur a cost, and each task is attached a weight to indicate its importance. Our objective is to determine a recruiting strategy to maximize the total weighted completion quality under a limited budget. We model such an unknown worker recruitment process as a novel combinatorial multi-armed bandit problem, and propose an extended UCB based worker recruitment algorithm. Moreover, we extend the problem to the case where the workers' costs are also unknown and design the corresponding algorithm. We analyze the regrets of the two proposed algorithms and demonstrate their performance through extensive simulations on real-world traces.

Mobile CrowdSensing (MCS) is a promising paradigm that recruits users to cooperatively perform various sensing tasks. In most realistic scenarios, users dynamically participate in MCS, and hence, we should recruit them in an online manner. In general, we prefer to recruit a user who can make the maximum contribution at the least cost, especially when the recruitment budget is limited. The existing strategies usually formulate the user recruitment as the budgeted optimal stopping problem, while we argue that not only the budget but also the time constraints can greatly influence the recruitment performance. For example, if we have less remaining budget but plenty of time, we should recruit users with more patience. In this paper, we propose a dynamic user recruitment strategy with truthful pricing to address the online recruitment problem under the budget and time constraints. To deal with the two constraints, we first estimate the number of users to be recruited and then recruit them in segments. Furthermore, to correct estimation errors and utilize newly obtained information, we dynamically re-adjust the recruiting strategy and also prove that the proposed strategy achieves a competitive ratio of (1 - 1/e)2/7. Finally, a reverse auction-based online pricing mechanism is lightly built into the proposed user recruitment strategy, which achieves truthfulness and individual rationality. Extensive experiments on three real-world data sets validate the proposed online user recruitment strategy, which can effectively improve the number of completed tasks under the budget and time constraints.

Real-time analytics on video data demands intensive computation resources and high energy consumption. Traditional cloud-based video analytics relies on large centralized clusters to ingest video streams. With edge computing, we can offload compute-intensive analysis tasks to the nearby server, thus mitigating long latency incurred by data transmission via wide area networks. When offloading frames from the front-end device to the edge server, the application configuration (frame sampling rate and frame resolution) will impact several metrics, such as energy consumption, analytics accuracy and user-perceived latency. In this paper, we study the configuration adaption and bandwidth allocation for multiple video streams, which are connected to the same edge node sharing an upload link. We propose an efficient online algorithm, called JCAB, which jointly optimizes configuration adaption and bandwidth allocation to address a number of key challenges in edge-based video analytics systems, including edge capacity limitation, unknown network variation, intrusive dynamics of video contents. Our algorithm is developed based on Lyapunov optimization and Markov approximation, works online without requiring future information, and achieves a provable performance bound. Simulation results show that JCAB can effectively balance the analytics accuracy and energy consumption while keeping low system latency.

Network Function Virtualization (NFV) transforms the implementation of network functions from expensive hardwares to software middleboxes. These software middleboxes, also called Virtual Network Functions (VNFs), are executed on virtualization platforms, which makes them more prone to error compared to dedicated hardwares. One effective way of ensuring VNF robustness is to provision redundancy in the form of deploying backup instances. Flows usually request to be processed by a service chain, consisting of multiple chained VNFs in some order. This paper considers both the resilient VNF deployment and the routing of flows. We deploy both active and backup VNF instances while guaranteeing the required VNF service resilience. Our objective is to minimize the total expected transmission delay of all flows when the probabilistic prior failure information is given. We first formulate our problem mathematically. We discuss the solution for the general network setting. Then, we simplify some setups and propose performance-guaranteed solutions in various scenarios with detailed approximation ratio analysis: for the case of a single flow without backup instances, we propose four optimal solutions corresponding to the settings of link transmission delay and server capacity; for the case of a single flow with backup instances, we extend the algorithms; for the case of two flows without backup instances, we propose efficient algorithms with approximation ratios inspired by the solutions for a single flow.

As data center clusters need to process quintillion bytes of data per day, it becomes a critical problem that efficiently scheduling jobs to improve resource utilization. However, the data analysis job usually contains multiple stages with dependent relationships, which brings challenges for scheduling. Those stages are modeled as Directed Acyclic Graphs (DAGs) and the general DAG scheduling problem is NP-hard. In this paper, we notice that in some parallel computing frameworks such as Spark, the execution of each stage could be divided into multiple phases that use different resources. We observe that interleaving different resources in a pipelined manner could improve resource utilization. Based on this observation, we propose to minimize the job makespan by exploiting resource pipeline. We first theoretically analyze the scheduling for perfectly parallel stages. In this case, our scheduling problem is equivalent to a DAG shop problem which is NP-hard. A contention-free scheduler is proposed and its approximation properties are analyzed. Stages of real-world jobs are usually not perfectly parallel. For general jobs, a reinforcement learning (RL) based scheduler is proposed to adaptively adjust the resource contention. We evaluate our contention-free and RL-based schedulers on a Spark cluster deployed on the Amazon EC2. Experiments on real-world and synthetic datasets show our RL-based scheduler can improve the CPU and network utilization by 33.0% and 29.7%, respectively.

Besides the computation limitation, the requirement of storing the entire blockchain is another challenge for blockchain mining in mobile environments, and thus has hindered the development of blockchain-powered mobile applications. Storage outsourcing to a cloud service provider (CSP) is a viable solution. An individual miner can store his blockchain in the cloud and then validate transactions by querying the CSP. However, validation outsourcing to a remote CSP incurs delay and damages a miner's winning probability in the mining competitions. To shorten such an unwanted delay, miners can also cache the unspent transaction output (UTXO) set in a nearby edge service provider (ESP) for fast transaction validations, which definitely brings extra costs. In this paper, we consider a two-layer outsourcing paradigm to solve storage shortage for mobile miners. Due to the delay-cost tradeoff when selecting service providers, we can model interactions among miners as a non-cooperative game and formulate a Nash equilibrium problem to investigate the effects of outsourcing on miners' utilities. We also study the access probability of UTXOs with different generation times. This will guide miners on how to select unspent transaction outputs if they decide only to cache the partial UTXO set in the edge. We further extend our game by modeling multiple mining rounds as a one-shot game to see how the cache update frequency affects miners' strategies. Numerical evaluation is conducted to show the feasibility of storage outsourcing and to validate the proposed models and theoretical results.

Edge caching has been emerged as a promising solution to alleviate the redundant traffic and the content access latency in the future Internet of Vehicles (IoVs). Several Reinforcement Learning (RL) based edge caching methods have been proposed to improve the cache utilization and reduce the backhaul traffic load. However, they can only obtain the local sub-optimal solution, as they neglect the influence of environment by other agents. In this paper, we investigate the edge caching strategy with consideration of the content delivery and cache replacement by exploiting the distributed Multi-Agent Reinforcement Learning (MARL). We first propose a hierarchical edge caching architecture for IoVs and formulate the corresponding problem with the objective to minimize the long-term cost of content delivery in the system. Then, we extend the Markov Decision Process (MDP) in the single agent RL to the multi-agent system, and propose a distributed MARL based edge caching algorithm to tackle the optimization problem. Finally, extensive simulations are conducted to evaluate the performance of the proposed distributed MARL based edge caching method. The simulation results show that the proposed MARL based edge caching method significantly outperforms other benchmark methods in terms of the total content access cost, edge hit rate and average delay. Especially, our proposed method greatly reduces an average of 32% total content access cost compared with the conventional RL based edge caching methods.

Mobile CrowdSensing (MCS) has recently emerged as a practical paradigm for large-scale and fine-grained urban sensing systems. To reduce sensing cost, Sparse MCS only senses data from a few subareas instead of sensing the full map, while the other unsensed subareas could be inferred by the intradata correlations among the sensed data. In certain applications, users are not only interested in inferring the data of other unsensed subareas in the current sensing cycle, but also interested in predicting the full map data of the near future sensing cycles. However, the intradata correlations exploited from the historical sparse sensed data cannot be effectively used for predicting full data in the temporal-spatial domain. To address this problem, in this paper, we propose an urban prediction scheme via Sparse MCS consisting of the matrix completion and the near-future prediction. To effectively utilize the sparse sensed data for prediction, we first present a bipartite-graph-based matrix completion algorithm with temporal-spatial constraints to accurately recover the unsensed data and preserve the temporal-spatial correlations. Then, for predicting the fine-grained future sensing map, with the historical full sensing data, we further propose a neural-network-based continuous conditional random field, including a Long Short-Term Memory component to learn the non-linear temporal relationships, and a Stacked Denoising Auto-Encoder component to learn the pairwise spatial correlations. Extensive experiments have been conducted on three real-world urban sensing data sets consisting of five typical sensing tasks, which verify the effectiveness of our proposed algorithms in improving the prediction accuracy with the sparse sensed data.

With the success of voice recognition techniques, users can easily control any device in smart home environments by simply saying a voice command. Based on this idea, a new group of smart devices are designed and released, which are called voice assistant. However, the voice itself is not secure and can be attacked in many ways. To defend against various types of voice replay attacks, we present a new voice liveness detection system. The basic insight of our system is that mouth opening movements will change the space size in the ear canal, which further changes the air pressure in ear canals. In this paper, we propose solutions to detect mouth opening movements using the noisy air pressure data and match them with the voices to validate the liveness of the voice source. To evaluate the effectiveness of our system, we develop a prototype on Raspberry Pi and conduct comprehensive evaluations. Experiments with ten volunteers show that our system can accurately accept voice commands from legitimate users with an accuracy of 91.72%. Moreover, our system can effectively defend current voice assistant devices from replay attacks with an accuracy of 97.2%.

Proof of Capacity (PoC) is an eco-friendly alternative to Proof of Work for consensus in blockchains since it determines mining rights based on miners' storage rather than computation. In PoC, for every block, a miner executes hashing on part of his dedicated storage. The miner that comes up with the smallest hash value among all miners will win the block. PoC has yet to be applied to mobile applications, due to the storage limitation of mobile devices. Storage offloading can be a viable solution that allows miners to offload mining all files to a cloud storage. In each mining round, a miner can decide whether to mine on his local device or by a cloud virtual machine (VM). Self-mining requires no extra cost but it incurs download delay, which will reduce the chance of winning. Cloud-mining experiences no delay but it brings cost on VMs. This delay-cost tradeoff challenges each miner to determine a ratio between self-mining and cloud-mining to maximize his utility. We model interactions among miners as a non-cooperative game and formulate a Nash equilibrium problem to investigate the effects of offloading on miners' utilities. We analyze the existence and uniqueness of equilibrium and propose a distributed algorithm to achieve the equilibrium in a uniform-delay setting. Further, we extend our results to non-uniform delays since miners may choose different network settings, e.g. 5G, 4G, or 3G. Both numerical evaluation and testbed experiments on Burstcoin are conducted to show the feasibility of storage offloading and to validate the proposed models and theoretical results.

Submodularity is an important property of set functions and has been extensively studied in the literature. It models set functions that exhibit a diminishing returns property, where the marginal value of adding an element to a set decreases as the set expands. This notion has been generalized to considering sequence functions, where the order of adding elements plays a crucial role and determines the function value; the generalized notion is called sequence (or string) submodularity. In this paper, we study a new problem of robust sequence submodular maximization with cardinality constraints. The robustness is against the removal of a subset of elements in the selected sequence (e.g., due to malfunctions or adversarial attacks). Compared to robust submodular maximization for set function, new challenges arise when sequence functions are concerned. Specifically, there are multiple definitions of submodularity for sequence functions, which exhibit subtle yet critical differences. Another challenge comes from two directions of monotonicity: forward monotonicity and backward monotonicity, both of which are important to proving performance guarantees. To address these unique challenges, we design two robust greedy algorithms: while one algorithm achieves a constant approximation ratio but is robust only against the removal of a subset of contiguous elements, the other is robust against the removal of an arbitrary subset of the selected elements but requires a stronger assumption and achieves an approximation ratio that depends on the number of the removed elements. Finally, we generalize the analyses to considering sequence functions under weaker assumptions based on approximate versions of sequence submodularity and backward monotonicity.

Nowadays, personal identification number (PIN) is one of the most popular methods for identity verification. However, recent researches show that attackers can easily recover victims’ PINs in spite of the large number of combinations PIN provides. Existing protection approaches require alteration of the original interaction between the user and PIN-based authentication systems, or still fail if the attacker can observe and mimic the victim’s input behavior. Considering these limitations, we propose a defense system called LightDefender to protect current PIN-based systems from PIN replay attacks using a single ambient light sensor. Specifically, we protect the PIN input by leveraging the biometrics in the received light intensity that is influenced by input behaviors and biological features. To our best knowledge, our work is the first one to protect PIN input using the light intensity. Different from existing approaches, LightDefender does not change the original interaction methods between the user and PIN-based authentication systems, and the extra hardware cost is low. In addition, by leveraging biological differences (e.g. finger length) among different users, LightDefender still claims high-security protection against strong attackers who can mimic the victim’s input behaviors. Experiments with 10 volunteers show that LightDefender can achieve an average true acceptance rate of 95% for normal users. More importantly, LightDefender can correctly reject two types attackers with an average true rejection rate of at least 93.6% without data of new attackers.

Mobile Crowdsensing (MCS) is a new paradigm that recruits users to cooperatively perform a sensing task. When recruiting users, existing works mainly focus on selecting a group of users with the best objective ability, e.g., the user's probability or frequency of covering the task locations. However, we argue that, for the cooperative MCS task, the completion effect depends not only on the user's objective ability, but also on their subjective collaboration likelihood with each other. In other words, in each single round, we prefer to recruit users with not only a strong objective ability but also good collaboration likelihood. Moreover, even though we could find a well-behaved group of users in a single round, in the multi-round scenario without enough prior knowledge, we still face the problem of recruiting previously well-behaved user groups (exploitation) or recruiting unknown user groups (exploration). To address these problems, in this paper, we first convert the single-round user recruitment problem into the min-cut problem and propose a graph theory based algorithm to find the optimal group of users. Furthermore, in the multi-round scenario, to balance the trade-off between exploration and exploitation, we propose the multi-round User Recruitment strategy based on the combinatorial Multi-armed Bandit model (URMB) and prove that it can achieve a tight regret bound. Finally, extensive experiments on three real-world datasets validate that the users recruited by URMB result in a better task completion effect than the state-of-the-art strategy.

Serendipity recommendation has attracted more and more attention in recent years; it is committed to providing recommendations which could not only cater to users’ demands but also broaden their horizons. However, existing approaches usually measure user-item relevance with a scalar instead of a vector, ignoring user preference direction, which increases the risk of unrelated recommendations. In addition, reasonable explanations increase users’ trust and acceptance, but there is no work to provide explanations for serendipitous recommendations. To address these limitations, we propose a Directional and Explainable Serendipity Recommendation method named DESR. Specifically, we extract users’ long-term preferences with an unsupervised method based on GMM (Gaussian Mixture Model) and capture their short-term demands with the capsule network at first. Then, we propose the serendipity vector to combine long-term preferences with short-term demands and generate directionally serendipitous recommendations with it. Finally, a back-routing scheme is exploited to offer explanations. Extensive experiments on real-world datasets show that DESR could effectively improve the serendipity and explainability, and give impetus to the diversity, compared with existing serendipity-based methods.

Along with the generation of Internet of Things (IoT), the values of tremendous volumes of sensing data will be slowly unlocked. Thus, crowd-sensed data trading as a new business paradigm has recently attracted increasing attention. A typical data trading system contains a platform, data consumers, and crowd workers. The platform recruits crowd workers to collect data and then sells the data to consumers. In this article, we design a differentially private crowd-sensed data trading mechanism, called DPDT, to preserve the identity privacy of consumers and the task privacy against crowd workers during the data collection process, simultaneously. DPDT consists of a differentially private auction-based data pricing algorithm and a differentially private data collection algorithm. The data pricing algorithm achieves a good approximation to the maximum revenue. Meanwhile, it guarantees (e2- 1)ϵ-truthfulness and 2ϵ-differential privacy, where ϵ > 0 is a small constant. The data collection algorithm is able to effectively protect the data collection task privacy against crowd workers. We prove that this data collection algorithm achieves δ-approximate ϵ-differential privacy, where δ <; 1/e is a small constant, and meanwhile guarantees a tight bound of the expected approximation ratio. At last, extensive simulations are conducted to verify the significant performance of DPDT.

Sparse mobile crowdsensing is a practical paradigm for large sensing systems, which recruits a small number of users to sense data from only a few subareas and, then, infers the data of unsensed subareas. In order to provide high-quality sensing services under a budget constraint, we would like to select the most effective users to collect useful sensing data to achieve the highest inference accuracy. However, due to the variable user mobility and complicated data inference, it is really challenging to directly select the best user set which helps the most with data inference. From the user's side, we can obtain the probabilistic coverage according to the users' mobilities, while the probabilistic coverage cannot indicate the data inference accuracy directly. From the subarea's side, we may identify some more useful subareas under the current states (e.g., the previous sensed subareas and the current expected coverage), while these useful subareas may not be covered by the users. Moreover, both the user mobility and data inference introduce a lot of uncertainty, which yields nonmonotonicity and thus nonsubmodularity in the user recruitment problem. Therefore, in this article, we study the user recruitment problem on both the user's and subarea's sides and propose a three-step strategy, including user selection, subarea selection, and user-subarea-cross (US-cross) selection. We first select some candidate user sets, which may cover the most subareas under the budget constraint (user selection), then estimate which subareas are more useful on data inference according to the selected candidates (subarea selection), which finally guides us to recruit the best user set (US-cross selection). Extensive experiments on two real-world data sets with four types of sensing tasks verify the effectiveness of our proposed user recruitment algorithms, which can effectively enhance the data inference accuracy under a budget constraint.

Internet of Things (IoT) devices and the edge jointly broaden the IoT's sensing capability and the monitoring scope for various applications. Though accessing sensing data and making decisions through IoT smart devices turns out to be commonplace, it is challenging to guarantee user privacy and preserve the accuracy (integrity) of the collected data. The IoT smart devices frequently lose either IoT user's privacy or data integrity. This also makes it crucial to put a threshold on the cost of computation and load of the IoT devices, as gradually more IoT services demand access to the resources that devices offer. In this article, we propose BalancePIC, a scheme that attempts to preserve a balance in the three aspects (user privacy, data integrity in edge-assisted IoT devices, and the computational cost). It achieves the balance through a balanced truth discovery approach and a proposed enhanced technique for data privacy, which are used in IoT devices and edge server interactions. It authenticates the IoT user participation with privacy in the truth discovery process through a biometric-ECC-based authentication algorithm. The nature of the BalancePIC scheme is to straightforwardly provide the likelihood for a simple amendment on the cryptography technique and weight assignment. This lessens the overall computational cost for the IoT user devices but also restricts the communications between the user devices and the edge server, which is important for data integrity. We present an enhanced technique to preserve privacy by guarding the user from potential threats and suspicious data collection parties. To achieve this, BalancePIC takes steps to blur the original sensory data of the device by processing results in groups called zones. Simulation result analysis provides evidence for the balance preservation in the three aspects.

Edge computing is envisioned to be a powerful platform that provides efficient data storage and computation services in the smart Internet-of-Things (IoT) systems. In this data-intensive architecture, protecting user-side data privacy is one of the most critical concerns to prevent privacy leakage from any other untrusted entities. Aiming to resist this concern, many privacy-preserving data aggregation (PPDA) schemes have been proposed for various cloud-enabled IoT applications. However, due to the resource-constrained nature of the smart IoT devices, the conventional PPDA solutions, in terms of both privacy and performance requirements, are unsuitable in edge computing. To address this challenge, we propose a lightweight and verifiable PPDA scheme, named LVPDA, for the edge-computing-enabled IoT system, where the Paillier homomorphic encryption method and an online/offline signature technique are combined to ensure the privacy preserving and integrity verification during the data aggregation process. A detailed security analysis indicates that LVPDA is existentially unforgeable under the chosen message attack (EU-CMA) and the data integrity can be guaranteed with formal proof under q -strong Diffie-Hellman (q -SDH) assumptions. Compared with other PPDA methods, our scheme can achieve lightweight PPDA in terms of less computational complexity and communication overhead.

Indoor localization is of great significance to a wide range of applications in the era of mobile computing. The maturity of the computer vision techniques and the ubiquity of embedded sensors in commercial off-the-shelf (COTS) smartphones shed the light on the submeter localization services for indoor environment. The state-of-the-art indoor localization works suffer from high-cost deployment and inaccurate results due to the coarse readings from internal measurement units (IMUs) sensors in the smartphones. In this article, we mainly innovate in introducing the WiFi-sensing technology to extract the distance information in a low-cost and device-free manner. Along with the computer vision technology, we model and implement an accurate and easy-to-deploy system for indoor localization. This system enhances indoor localization with multimodal sensing via two images, IMU sensors reading and CSI of WiFi signal. Specifically, we first model and design camera-based, sensor and WiFi-assisted indoor localization and propose several algorithms in this model. We then implement a prototype with smartphones and commercial WiFi devices and evaluate it in several distinct indoor environments. The experimental results show that 92-percentile error is within 0.2 m for indoor targets which sheds light on submeter indoor localization.

Offloading cellular traffic through WiFi Access Points (APs) has been a promising way to relieve the overload of cellular networks. However, data offloading process consumes a lot of resources (e.g., energy, bandwidth, etc.). Given that the owners of APs are rational and selfish, they will not participate in the data offloading process without receiving the proper reward. Hence, there is an urgent need to develop an effective incentive mechanism to stimulate APs to take part in the data offloading process. This paper proposes a novel Delay-constraint and Reverse Auction-based Incentive Mechanism, named DRAIM. In DRAIM, we model the reverse auction-based incentive problem as a nonlinear integer problem from the business perspective, aiming to maximize the revenue of the Mobile Network Operator (MNO), and jointly consider the delay constraint of different applications in the optimization problem. Then, two low-complexity methods: Greedy Winner Selection Method (GWSM), and Dynamic Programming Winner Selection Method (DPWSM) are proposed to solve the optimization problem. Furthermore, an innovative standard Vickrey-Clarke-Groves scheme-based payment rule is proposed to guarantee the individual rationality and truthfulness properties of DPWSM. At last, extensive simulation results show that the proposed DPWSM is superior to the proposed GWSM and the Random Winner Selection Method in terms of the MNO’s utility and traffic load under different scenarios.

Extracting knowledge from sensor data for various purposes has received a great deal of attention by the data mining community. For the purpose of event detection in cyber-physical systems (CPS), e.g., damage in building or aerospace vehicles from the continuous arriving data is challenging due to the detection quality. Traditional data mining schemes are used to reduce data that often use metrics, association rules, and binary values for frequent patterns as indicators for finding interesting knowledge about an event. However, these may not be directly applicable to the network due to certain constraints (communication, computation, bandwidth). We discover that, the indicators may not reveal meaningful information for event detection in practice. In this paper, we propose a comprehensive data mining framework for event detection in the CPS named DPminer, which functions in a distributed and parallel manner (data in a partitioned database processed by one or more sensor processors) and is able to extract a pattern of sensors that may have event information with a low communication cost. To achieve this, we introduce a new sensor behavioral pattern mining technique called differential sensor pattern (DSP) which considers different frequencies and values (non-binary) with a set of sensors, instead of traditional binary patterns. We present an algorithm for data preparation and then use a highly-compact data tree structure (called DP-Tree) for generating the DSP. An important tradeoff between the communication and computation costs for the event detection via data mining is made. Evaluation results show that DPminer can be very useful for networked sensing with a superior performance in terms of communication cost and event detection quality compared to existing data mining schemes.

Mobile crowdsensing recruits a group of mobile users to cooperatively perform a common sensing job with their smart devices. As a special issue, photo crowdsensing allows users to utilize the built-in cameras of mobile devices to take photos for an event or a target. Then, the photos can be used in numerous application areas, such as target reconstruction, scenario reduction, and so on. Therefore, photo crowdsensing has attracted considerable attention recently due to the rich information that can be provided by images. In this paper, we focus on using the photos to make reconstructions for specific targets. Furthermore, we develop a user recruitment system for efficient photo collecting in mobile crowdsensing (RSMC), where the task requesters publish a sensing task to the users, and the map is gridded according to the locations of the sensing targets. Then, we use a semi-Markov model to calculate the user's utility for the sensing task. Finally, a user recruitment strategy is devised to recruit the optimal k users for finishing the sensing task. We conduct extensive simulations based on three widely used real-world traces: roma/taxi, epfl, and geolife. The results show that, compared with other recruitment strategies, RSMC takes the largest number of efficient photos for the sensing task.

In this article, an industrial cyber-physical system (ICPS) is utilized for monitoring critical events such as structural equipment conditions in industrial environments. Such a system can easily be a point of attraction for the cyberattackers, in addition to system faults, severe resource constraints (e.g., bandwidth and energy), and environmental problems. This makes data collection in the ICPS untrustworthy, even the data are altered after the data forwarding. Without validating this before data aggregation, detection of an event through the aggregation in the ICPS can be difficult. This article introduces TrustData, a scheme for high-quality data collection for event detection in the ICPS, referred to as “Trust worthy and secured Data collection” scheme. It alleviates authentic data for accumulation at groups of sensor devices in the ICPS. Based on the application requirements, a reduced quantity of data is delivered to an upstream node, say, a cluster head. We consider that these data might have sensitive information, which is vulnerable to being altered before/after transmission. The contribution of this article is threefold. First, we provide the concept of TrustData to verify whether or not the acquired data are trustworthy (unaltered) before transmission, and whether or not the transmitted data are secured (data privacy is preserved) before aggregation. Second, we utilize a general measurement model that helps to verify acquired signal untrustworthy before transmitting toward upstream nodes. Finally, we provide an extensive performance analysis through a real-world dataset, and our results prove the effectiveness of TrustData.

This paper studies the practically important problem of data collection for sensor-augmented RFID systems. However, existing RFID data collection protocols suffer from two common limitations: execution time is naturally in proportion to the number of tags, thus they cannot satisfy time-stringent application scenarios; none of them is complaint with the C1G2 standard, thus they cannot be implemented using Commercial-Off-The-Shelf (COTS) RFID tags. To overcome these two limitations, this paper proposes the Geographical correlation-based RF-data Collection (GRC) protocol. GRC is fast because it is able to approximately capture the sensing data of all tags by only actually gathering data from a small set of sampled tags. This is based on the observation from the real-world data set that sensing data has a strong geographical correlation, i.e., data gathered from nearby RFID tags has similar values. In GRC, we use a greedy approach to find the minimum sampling tag set to cover the whole monitoring region such that each un-sampled tag has at least one sampled tag nearby. Then, RFID reader runs the Framed Slotted Aloha (FSA) protocol specified in C1G2 standard to collect sensing data from the sampled tags. For each un-sampled tag, we approximate its sensing data by calculating weight-average of the data collected from its nearby sampled tags, where a faraway sampled tag should be given a small weight, and vice versa. Compared with existing RFID data collection schemes, the advantages of GRC are two-fold: (1) Extensive simulation results demonstrate that the time cost of our GRC scheme is only 1/28~1/3 of the state-of-the-art data collection scheme; (2) GRC is totally complaint with C1G2 standard, thus it can be easily deployed on the COTS RFID tags.

Tag inventory is one of the most fundamental tasks for RFID systems. However, the Framed Slotted Aloha (FSA) protocol specified in the C1G2 standard is of low time-efficiency, because it needs to collect all tags in the system. To improve time-efficiency, research communities proposed a batch of sampling-based approaches, in which the reader only needs to collect a small set of sampled tags instead of all. Although time-efficiency has been improved, existing sampling-based approaches still have two common limitations. First, all tags in the system are assumed to have the same sampling probability. It is unfair that tags attached to differential items (e.g., different values) have the same chance to be sampled and collected. Second, all existing sampling-based approaches stay in theory level and cannot be deployed on Commercial Off-The-Shelf (COTS) RFID devices, because the C1G2 standard does not support the sampling function at all. To deal with the above two limitations, this paper studies the new problem of differential tag sampling-letting each RFID tag be identified with a given sampling probability. In this paper, we use the COTS RFID devices including Impinj Speedway R420 reader and Monza 4QT tags to implement the Differential Tag Sampling (DTS) operation. Then, we apply probabilistic analytics on the collected tag data to address some practically important problems such as Multi-category Tag Cardinality Estimation (MTCE), and Value-based Missing Tag Detection (VMTD). Although the analytics results are not 100 percent accurate, the deviation in the results can be controlled below a small threshold and DTS can significantly improve the time-efficiency. DTS can be easily deployed on the COTS RFID systems, because it is totally compliant with the C1G2 standard. Extensive experiments demonstrate that DTS is able to let each tag take the given sampling probability to be sampled and identified. Moreover, the proposed DTS protocol can significantly reduce the execution time of MTCE and VMTD by nearly 70 percent than the FSA protocol.

Network Function Virtualization (NFV) has promoted the implementation of network functions from expensive hardwares to software middleboxes. These software middleboxes, also called Virtual Network Functions (VNFs), are executed on switch-connected servers. Efficiently deploying such VNFs on servers is challenging because the traffic rate of flows must be fully processed by their requested VNFs when they reach destinations, and the deployed positions of VNFs are restricted by the server capacity. In addition, each network function offers non-uniform VNF models (types) with different configurations of processing volumes and costs. This paper focuses on minimizing the total cost of deploying VNFs for providing a specific network function to all flows in tree-structured networks. First, we prove the NP-hardness of non-uniform VNF deployment in a tree topology and propose a dynamic programming based solution with a pseudo-polynomial time complexity. Then we narrow it down to three simplified cases by focusing on either uniform VNFs or the linear line topology. Specifically, three algorithms are introduced: an improved dynamic programming based algorithm for deploying uniform VNFs in a tree topology, a performance-guaranteed algorithm for deploying non-uniform VNFs in a linear line topology, and an optimal greedy algorithm for deploying uniform VNFs in a linear line topology. Additionally, we generalize our approach to a case of deploying a service chain, which consists of multiple network functions applied to flows in a specific order. We propose two solutions: one is optimal but time-consuming while another is heuristic but efficient. Extensive simulations are conducted to evaluate our algorithms.

Changes in network state are a common source of instability in networks. An update event typically involves multiple flows that compete for network resources at the cost of rescheduling and migrating some existing flows. Previous network updating schemes tackle such flows independently, rather than as the entity of an update event. They only optimize the flow-level metrics for the flows involved in an update event. In this paper, we present an event-level abstraction of network update that groups flows of an update event and schedules them together to minimize the event completion time (ECT). We then study the scheduling problem of multiple update events for achieving high scheduling efficiency and preserving fairness. The designed least migration traffic first (LMTF) method schedules all update events in the FIFO order, but it avoids head-of-line blocking by randomly fine-tuning the queue order of some events. It can considerably reduce the update cost, the average, and tail ECTs of update events. In addition, we design a general parallel-LMTF (P-LMTF) method to guarantee fairness and further improve scheduling efficiency among update events. This improves the LMTF method by opportunistically updating multiple events simultaneously. The comprehensive evaluation results indicate that the average ECT of our approach is up to 10× faster than the flow-level scheduling method for network update events, and its tail ECT is up to 6× faster. Our P-LMTF method incurs a 75% reduction in the average ECT compared with FIFO when the network utilization exceeds 70%, and it achieves a 42% reduction in tail ECT.

Product reviews play an important role in guiding users’ purchase decision-making in e-commerce platforms. However, it is challenging for users to find helpful reviews that meet their preferences and experiences among an overwhelming amount of reviews. Some works have been done to recommend helpful reviews to users, either from personalized or non-personalized views. While some existing models recommend similar users’ reviews for a target user, they either neglect the target user’s aspect preferences or the user-product interactions for measuring user similarity. Moreover, those models predict review helpfulness at the review-level (a review is taken as a whole); few of them consider the aspect-level. To address the above issues, we propose an aspect sentiment similarity-based personalized review recommendation model ( A2SPR ), which quantifies review helpfulness and recommends reviews that are customized for each individual. We analyze users’ aspect preferences from reviews and improve user similarity with users’ fine-grained sentiment and product relevance. Furthermore, we redefine the review helpfulness score at the aspect level, which indicates the review’s reference value for users’ purchase decisions. Finally, we recommend the top k helpful reviews for individuals based on the review helpfulness score. To validate the performance of the proposed model, eight baselines are developed and compared. Experimental results show that our model performs better than those baselines in both the coverage and precision.

Mobile Augmented Reality (MAR) applications usually contain computation-intensive tasks which far outstrip the capability of mobile devices. One way to overcome this is offloading computation-intensive MAR tasks to remote clouds. However, the wide area network delay is hard to reduce. Thanks to edge computing, we can offload MAR tasks to nearby servers. Prior studies focus on either single-task MAR applications offloading or dependent tasks offloading for a single user. In this article, we study the offloading decision of MAR applications from multiple users, each of which is comprised of a chain of dependent tasks, over a generic cloud-edge system consisting of a group of heterogeneous edge servers and remote clouds. We formulate the Multi-user Multi-task MAR Application Scheduling (M3AS) problem, which is NP-hard. We present Mutas, an efficient scheduling algorithm that jointly optimizes server assignment and resource management. We also consider the online version of M3AS and present OnMutas. Extensive evaluations demonstrate that both Mutas and OnMutas can significantly reduce the service delays of MAR applications when compared to three other heuristics.

Mobile crowdsensing is a new paradigm in which a requester can recruit a group of mobile users via a platform and coordinate them to perform some sensing tasks by using their smartphones. In mobile crowdsensing, each user might perform multiple tasks with different sensing qualities. Meanwhile, the users participating in the crowdsensing will ask for sufficient rewards to compensate for their expenditures. Hence, an important problem is how to recruit the users with minimum cost while achieving a satisfactory sensing quality for each task. Furthermore, in order to ease users' worries about privacy disclosures, the user recruitment process needs to protect each user's sensing quality and recruitment cost information from being revealed to other users or to the platform. In this paper, we propose two secure user recruitment problems for the cases where the recruitment costs of users are homogeneous and heterogeneous. After proving the NP-hardness of the problems, we design two secure user recruitment protocols by using secret sharing scheme. Both of the proposed protocols adopt greedy strategies, which can recruit nearly optimal users while ensuring that the total sensing quality of each task is no less than a given threshold. The difference lies in that the two greedy strategies are based on two unique utility functions. We analyze the approximation ratios of the two protocols and prove the security under the semi-honest model. Finally, we demonstrate the significant performance of the proposed protocols through extensive simulations and executions on real smartphones.

Virtualization techniques help edge environments separate the role of the traditional edge providers into two: edge infrastructure providers (EIPs), who manage the physical edge infrastructure, and edge service providers (ESPs), who aggregate resources (especially, compute resources) from multiple EIPs to place service entities and offer value-added services to end users (EUs). In such an environment, end users submit their data analysis jobs to ESPs; ESPs process the data analysis jobs using their service entities. One fundamental and critical problem for an ESP is to decide how much compute resources to rent from each edge server under the constraint that the total amount of rental resources is no more than a specified budget threshold, so that the average makespan of the data analysis jobs submitted to it is minimized. This Edge Resource Allocation (ERA) problem is proven to be NP-complete by reducing the set cover problem to a special case of it. To design an approximation algorithm for ERA, we perform two transformations on ERA: first, we transform ERA into mERA by replacing minimization with maximization; second, we transform mERA into dmERA by limiting the possible amounts of rental resources to a finite set of values. We find that dmERA has several tractable properties that allow us to design Hermes, a provably efficient algorithm that approximates the optimal allocation. We demonstrate that the gap between Hermes and the optimum in simulations and Android-based testbed experiments are no larger than 4.78% and 12.43%, respectively. Hermes can also output a curve showing the trade-off between the average makespan and the budget threshold, so that an ESP can choose the right balance.

Mobile Edge Computing (MEC) reforms the cloud paradigm by bringing unprecedented computing capacity to the vicinity of end users at the mobile network edge. This provides end users with swift and powerful computing and storage capacities, energy efficiency, and mobility- and context-awareness support. Furthermore, Network Function Virtualization (NFV) is another promising technique that implements various network functions for many applications as pieces of software in servers or cloudlets in MEC networks. The provisioning of virtualized network services in MEC can improve user service experiences, simplify network service deployment, and ease network resource management. However, user requests arrive dynamically and different users demand different amounts of resources, while the resources in MEC are dynamically occupied or released by different services. It thus poses a significant challenge to optimize the performance of MEC through efficient computing and communication resource allocations to meet ever-growing resource demands of users. In this paper, we study NFV-enabled multicasting that is a fundamental routing problem in an MEC network, subject to resource capacities on both its cloudlets and links. Specifically, we first devise an approximation algorithm for the cost minimization problem of admitting a single NFV-enabled multicast request. We then develop an efficient algorithm for the throughput maximization problem for the admissions of a given set of NFV-enabled multicast requests. We third devise an online algorithm with a provable competitive ratio for the online throughput maximization problem when NFV-enabled multicast requests arrive one by one without the knowledge of future request arrivals. We finally evaluate the performance of the proposed algorithms through experimental simulations. Simulation results demonstrate that the proposed algorithms are promising.

Motivated by privacy preservation requirements for outsourced data, keyword searches over encrypted cloud data have become a hot topic. Compared to single-keyword exact searches, multi-keyword fuzzy search schemes attract more attention because of their improvements in search accuracy, typo tolerance, and user experience in general. However, existing multi-keyword fuzzy search solutions are not sufficiently efficient when the file set in the cloud is large. To address this, we propose an Efficient Leakage-resilient Multi-keyword Fuzzy Search (EliMFS) framework over encrypted cloud data. In this framework, a novel two-stage index structure is exploited to ensure that search time is independent of file set size. The multi-keyword fuzzy search function is achieved through a delicate design based on the Gram Counting Order, the Bloom filter, and the Locality-Sensitive Hashing. Furthermore, considering the leakages caused by the two-stage index structure, we propose two specific schemes to resist these potential attacks in different threat models. Extensive analysis and experiments show that our schemes are highly efficient and leakage-resilient.

Recommendation systems provide good guidance for users to find their favorite movies from an overwhelming amount of options. However, most systems excessively pursue the recommendation accuracy and give rise to over-specialization, which triggers the emergence of serendipity. Hence, serendipity recommendation has received more attention in recent years, facing three key challenges: subjectivity in the definition, the lack of data, and users' floating demands for serendipity. To address these challenges, we introduce a new model called HAES, a H ybrid A pproach for movie recommendation with E lastic S erendipity, to recommend serendipitous movies. Specifically, we (1) propose a more objective definition of serendipity, \em content difference and \em genre accuracy, according to the analysis on a real dataset, (2) propose a new algorithm named JohnsonMax to mitigate the data sparsity and build weak ties beneficial to finding serendipitous movies, and (3) define a novel concept of elasticity in the recommendation, to adjust the level of serendipity flexibly and reach a trade-off between accuracy and serendipity. Extensive experiments on real-world datasets show that HAES enhances the serendipity of recommendations while preserving recommendation quality, compared to several widely used methods.

Adversarial examples (AEs) are crafted by adding human-imperceptible perturbations to inputs such that a machine-learning based classifier incorrectly labels them. They have become a severe threat to the trustworthiness of machine learning. While AEs in the image domain have been well studied, audio AEs are less investigated. Recently, multiple techniques are proposed to generate audio AEs, which makes countermeasures against them urgent. Our experiments show that, given an audio AE, the transcription results by Automatic Speech Recognition (ASR) systems differ significantly (that is, poor transferability), as different ASR systems use different architectures, parameters, and training datasets. Based on this fact and inspired by Multiversion Programming, we propose a novel audio AE detection approach MVP-Ears, which utilizes the diverse off-the-shelf ASRs to determine whether an audio is an AE. We build the largest audio AE dataset to our knowledge, and the evaluation shows that the detection accuracy reaches 99.88%. While transferable audio AEs are difficult to generate at this moment, they may become a reality in future. We further adapt the idea above to proactively train the detection system for coping with transferable audio AEs. Thus, the proactive detection system is one giant step ahead of attackers working on transferable AEs.

A transit-link distributed denial-of-service (DDoS) attack is a special attack in which the attacker sends out a huge number of requests to exhaust the capacity of a link on the path the traffic comes to a server. As a result, denial-of-service and degradation of Quality-of-Service (QoS) occurs. Because the attack traffic does not go to the victim, protecting the legitimate traffic alone is hard for the victim. With the help of a special type of router called filter router (FR), the victim can protect the legitimate traffic. A FR can receive filter from servers and apply the filter to block a link incident to it. By analyzing traffic rates and paths, the victim can identify some links that may be congested. The victim needs to select some of these possible congested links and send a filter to the corresponding FR so that the legitimate traffic follows non-congested paths. In this paper, we formulate an optimization problem for selecting the minimum number of possible congested links so that the legitimate traffic goes through a non-congested path. We consider the scenario where every user has at least one non-congested shortest path. We transform the problem to the vertex separation problem to find the links to block. We build our own Java multi-threaded simulator and conduct extensive simulations.

One of the most prominent cellular technologies, Long Term Evolution (LTE), is currently operating on some 800MHz, 2GHz, and 3.5GHz licensed bands. Wi-Fi is currently operating on 2.5GHz and 5GHz unlicensed bands. The declaration stating that 5GHz bands are unlicensed enables LTE to operate on 5GHz bands. It is challenging, however, for different wireless technologies to co-exist. The two standards, LTE-U and LTE-LAA, for LTE to coexist with Wi-Fi on the 5GHz band have evolved. The LTE-U standard is based on the duty cycle, while LTE-LAA is based on listen-before-talk (LBT). In existing LTE-U systems, the LTE base station (eNB) estimates the fair portion of Wi-Fi usage based on channel state information. The usage estimation from channel state information is not as accurate enough as well as the fair portion. In this paper, we study the fair coexistence between LTE-U and Wi-Fi in the scenario where an LTE eNB can exchange information with Wi-Fi access points (AP). The communication can be done in both wired and wireless mediums. The wired medium is ethernet point-to-point communication, and the wireless communication is done using the reserved bits in Wi-Fi packets. Both ways are applicable to the operator, who has both LTE and Wi-Fi coverage. Therefore, the Wi-Fi AP can collect information about other APs and send it to its LTE eNB. The LTE eNB can adjust its parameters according to the received information to achieve fairness.

Cognitive radio networks (CRNs), offering novel network architecture for utilizing spectrum, have attracted significant attention in recent years. In CRNs, secondary users (SUs) first determine the status of a channel; if it is free, they start transmitting. If the status determination is wrong, SUs may unnecessarily interfere with the licensed primary user (PU). In cooperative spectrum sensing, a SU makes a decision about the presence of the PU based on its own and other SUs' sensing results. Malicious SUs (MSUs) send false sensing results to SUs so that they make wrong decisions about the PU presence. As a result, a SU may transmit during the presence of the PU or may keep starving for the spectrum. In this paper, we propose a reputation-based mechanism for cooperative spectrum sensing which can minimize the effects of MSUs on decision making. Some of the SUs are selected as distributed fusion centers (DFCs), which are responsible for making decisions about the PU presence and inform the reporting SUs. A DFC uses weighted majority voting among the reporting SUs, where weights are determined based on reputation. The DFC updates reputations of SUs based on confidence of an election. If the majority wins by a significant margin, the confidence of the election is high. In this case, SUs that belong to the majority get high reputations. We provide extensive simulations to validate our proposed model.

Ride-sharing companies such as Didi and Uber have served billions of passenger requests from all over the world. The efficiency of the ride-sharing is highly depended on the order dispatch system which assigns passenger requests to idle drivers. However, designing such a dispatch system is challenging because of the spatial-temporal dynamic of passenger requests, and the trade-off between the benefits for passengers and drivers. Existing order dispatch systems use either a system-assigning approach or a driver-grabbing approach. However, either approach has its own flaws. In this paper, we propose to combine the two existing approaches and jointly considers both passengers'' and drivers'' interest. In our approach, a passenger request is broadcast to the drivers in a dispatch region chosen by the system. The size of the dispatch region could iteratively increase until the request is accepted. We formulate an optimization problem to determine the increase speed of the dispatch region. Drivers'' idle driving distances and passengers'' waiting time are jointly considered. We propose a dynamic programming algorithm to optimally solve the increase ratio of the size of the dispatch region for a case that different dispatch regions are not overlapped. We further investigate the overlapped case and modify the dynamic programming algorithm correspondingly. We provide a discussion on the effect of the overlapping in a spatial case, where the driver and passenger locations are uniformly distributed. Experiments are conducted based on the synthetic dataset and the real-world dataset from Didi Inc. Results show that our approach can effectively reduce the expected driver pickup distance and keep the dispatching time short, which balances both passengers'' and drivers'' interests.

Lots of privacy and security issues in the current cloud-based data markets will be eliminated by taking advantage of blockchain-based decentralized storage services, which can provide a new paradigm for safe data outsourcing and correct remote search. However, existing data markets are also questioned on their inflexible and opaque pricing, where the value of data ownership and the cost of query search are mixed. Thus, a better pricing model is necessarily needed in an emerging decentralized data market. In this paper, we envision an Ethereum-based data market, in which the pricing model for each query includes two parties: owner (paid for his data ownership) and miner (rewarded by query search). We study a new cooperative search scheme through a proxy to reduce cost on the client (user) side. Suppose each user query is charged based on the number of keywords in the query. The cost reduction is based on combining multiple queries into a group subject to the constraint that the resulting combined query is not significantly larger than any of its original query in terms of the number of keywords. The total price is based on total number of keywords in all groups. As the optimal grouping depends on the pricing of both owner and miner, we build a small testbed to analyze how price setting will affect grouping results. Since it is a cooperative model with shared resources, we also study various incentive properties on the client side, thereby yielding a cost sharing mechanism to split joint cost in a truth-revealing and fair manner.

User dynamics in both spatial and temporal domains bring uncertainty to bike-sharing systems (BSSs) and usually lead to bike imbalance. This may generate out-of-service events due to bike underflow or overflow, at a bike station. In this paper, we recruit workers through crowdsourcing to rebalance loads among bike stations. We assume that workers have their individual sources and destinations, and assign them to move bikes from overflow stations to underflow stations. We partition the complex spatial and temporal problem into a sequence of slices with a fixed duration in the temporal domain. In each slice, we focus on the spatial domain and allocate a pair of overflow/underflow stations to a worker such that the summation of detour cost among workers is minimized. The hardness of finding the min-cost allocation is shown by a reduction from a 3-dimensional matching problem (i.e., matching among workers, overflow stations, and underflow stations). We propose a 3-approximation algorithm for the problem when the detour cost is proportional to the detour distances. Then, the configuration dynamic in the sequence of slices is captured by carefully determining the rebalancing frequency and target for each rebalancing operation. We investigate heuristic approaches to decide rebalancing frequencies and targets over a sequence of slices in order to minimize the total number of bike movements (i.e., the total number of workers), and hence to derive the average total detours per slice. We simulate our algorithms on both real-world and synthetic datasets based on different time-slice granularities. The experiment results show that our approaches can reduce the average total detour per slice.

Computation offloading has been considered as a viable solution to blockchain mining in mobile environments. In this paper, we present a two-layer computation offloading paradigm that includes an edge computing service provider (ESP) and a cloud computing service provider (CSP). We formulate a multi-leader multi-follower Stackelberg game to address the computing resource management problem in such a network, by jointly maximizing the profits of each service provider (SP) and the payoffs of individual miners. Two practical scenarios are investigated: a fixed-miner-number scenario for permissioned blockchains and a dynamic-miner-number scenario for permissionless blockchains. For the fixed-miner-number scenario, we discuss two different edge operation modes, i.e., the ESP is connected (to the CSP) or standalone, which form different miner subgames based on whether each miner's strategy set is mutually dependent. The existence and uniqueness of Stackelberg equilibrium (SE) in both modes are analyzed, according to which algorithms are proposed to achieve the corresponding SE(s). For the dynamic-miner-number scenario, we focus on the impact of population uncertainty and find that the uncertainty inflates the aggressiveness in the ESP resource purchasing. Numerical evaluations are presented to verify the proposed models.

Mobile Edge Computing (MEC) reforms the cloud paradigm by bringing unprecedented computing capacity to the vicinity of mobile users at the mobile network edge. This provides end-users with swift and powerful computing, energy efficiency, storage capacity, mobility-and context-awareness support. Furthermore, provisioning virtualized network services in MEC can improve user service experience, simplify network service deployments, and ease network resource management. However, user requests usually arrive into the system dynamically and different user requests may have different resource demands. How to optimize and guarantee the performance of MEC is of significant importance and challenging. In this paper, we study the problem of online NFV-enabled multicasting in an MEC network with resource capacity constraints on both cloudlets and links. We first devise an approximation algorithm for the cost minimization problem for a single NFV-enabled multicast request admission. We then propose an online algorithm with a provable competitive ratio for the online throughput maximization problem where NFV-enabled multicast requests arrive one by one without the knowledge of future request arrivals. We admit the requests through placing or sharing VNF instances of network functions in their service chains to meet their computing and bandwidth resource demands, and we introduce a novel cost model to capture the dynamic usages of different resources and perform network resource allocations based on the proposed cost model. We finally evaluate the performance of the proposed algorithms through experimental simulations. Simulation results demonstrate that the proposed algorithms are promising.

Using a mobile charger to wirelessly charge sensors is a promising yet not well-solved technique. Existing trajectory planning schemes for wireless charger either (1) fail to optimize the one-to-many characteristic of wireless charging or (2) fail to jointly optimize the charger movement cost and the charging cost. The objective of this paper is to find the optimal trajectory planning for a mobile charger in terms of energy minimization in the quadratic attenuation charging model. There exists a trade-off between charging efficiency and trajectory distance. If the mobile charger comes close to sensors, the charging efficiency is high, but the entire charging trajectory of the charger will be long and vice versa. To address this trade-off, we propose the idea of charging bundle and optimize the charger's trajectory based on the charging bundle rather than each sensor. The optimal charging bundle generation problem and the bundle trajectory optimization problem are discussed gradually. Both of them are proven to be NP-hard. Then, we first propose a greedy bundle generation algorithm with an approximation ratio of lnn, where n is the number of sensors. After that, we propose a TSP-based solution and further optimize the TSP-trajectory by jointly considering the adjacent charging locations. Theorems are proposed to effectively find the optimal location. Extensive experiments show that our scheme achieves a much better performance than traditional schemes.

The mission of set reconciliation (also called set synchronization) is to identify those elements which appear only in exactly one of two given sets. In this paper, we extend the set reconciliation problem into three design rationales: (i) multiset support; (ii) near 100% reconciliation accuracy; (iii) communication-friendly and time-saving. Prior reconciliation methods fail to realize the three rationales simultaneously. To this end, we redesign Trie and Fenwick Tree (FT), to near-accurately represent and reconcile two types of multisets that we refer to as unsorted and sorted multisets, respectively. Comprehensive evaluations are conducted to quantify the performance of our proposals. The trace-driven evaluations demonstrate that Trie and FT achieve near-accurate multiset reconciliation, with 4.31 and 2.96 times faster than the CBF-based method, respectively.

Computation offloading makes sense to the interaction between users and compute-intensive applications. Current researches focused on deciding locally or remotely executing an application, but ignored the specific offloading proportion of application. A full offloading cannot make the best use of client and server resources. In this paper, we propose an innovative reinforcement learning (RL) method to solve the proportional computation problem. We consider a common offloading scenario with time-variant bandwidth and heterogeneous devices, and the device generates applications constantly. For each application, the client has to choose locally or remotely executing this application, and determines the proportion to be offloaded. We formalize the problem as a long-term optimization problem, and then propose a RL-based algorithm to solve it. The basic idea is to estimate the benefit of posible decisions, of wihch the decision with the maximum benefit is selected. Instead of adopting the original Deep Q Network (DQN), we propose Advanced DQN (ADQN) by adding Priority Buffer Mechanism and Expert Buffer Mechanism, which improves the utilization of samples and overcomes the cold start problem, respectively. The experimental results show ADQN's high feasibility and efficiency compared with several traditional policies, such as None Offloading Policy, Random Offloading Policy, Link Capacity Optimal Policy, and Computing Capability Optimal Policy. At last, we analyse the effect of expert buffer size and learning rate on ADQN's performance.

Mobile crowdsensing, through which a requester can recruit a group of crowd workers via a platform and coordinate them to perform some sensing tasks, has attracted lots of attention recently. However, most of the existing mobile crowdsensing systems assume that the qualities of workers are known in advance. Based on this assumption, they study the task assignment and worker recruitment problems. Unfortunately, the qualities of workers are generally unknown in reality, so the platform must find the tradeoff between exploring and exploiting the qualities by using reinforcement learning. At the same time, all sensing tasks are required to be covered in each round (covering constraint), and the requester usually has a limited budget (budget constraint). In this paper, we study how to recruit unknown workers under the budget and covering constraints so that the total expected achieved qualities can be maximized. To this end, we model the problem as a combination of a maximum weight matching problem and a special multi-armed bandit problem. We first consider that the recruitment costs of workers are homogeneous and propose a recruitment algorithm with a performance guarantee. Then, we study the heterogenous case and devise a heuristic algorithm. Finally, we demonstrate the performances of our algorithms through extensive simulations.

The emerging Non-Volatile Memory (NVM) technology has given rise to an opportunity to accelerate big data analysis. In this paper, we investigate the joint job and data scheduling problem in private cloud data center with a hybrid storage system, and we propose Labeling Scheduler, a flexible labeling-based approach for jointly scheduling. The core idea of the approach is to introduce the labeling system to characterize the features of big data analysis jobs and data objects, and conduct data replacement dynamically between NVM and disk. To the best of our knowledge, this is the first work to introduce the labeling methodology to the big data analysis problem in the cloud data center with a hybrid storage system. We conduct extensive simulations and the simulation results show that the Labeling Scheduler has a significant improvement on system utility compared to the method without labeling information. In addition, the Labeling Scheduler guarantees a high NVM hit rate, which is valuable for NVM endurance enhancement.

Signal map is of great importance, especially in the dawn of 5G network, for site spectrum monitoring, location-based services (LBS), network construction, and cellular planning. Despite its significance, the traditional signal map construction, e.g., through full site survey, could be time-consuming and labor-intensive as the signal varies frequently over time and the accuracy requirement grows rapidly with the emergence of new applications. Even with crowdsourcing scheme, the participants tend to be unevenly distributed in space while the encouragement budgets for the participants could be far from enough to collect adequate high-quality measurements. Therefore, the signal map constructed by crowdsourcing is often sparse and incomplete. To this end, in this paper, we study how to effectively reconstruct and update the signal map in the case of partially measured signal maps with minimum cost and propose an auto-encoder-based active signal map reconstruction method (AER). Our method is mainly innovative in three parts. Firstly, AER can effectively update the signal map with only a small number of observations while also fully using the incomplete historical signals to effectively update the signal map online. Secondly, AER consists of an active query mechanism which quantitatively evaluates the most valuable measurement site for reconstruction, which further reduces the measurement cost to a large extent. Thirdly, to cope with the measurement dynamics, we give a new signal map model describing not only the signal strength but also the signal dynamics, based on which an advanced AER algorithm is proposed. The simulation results demonstrate the advantages and effectiveness of our approach in both accuracy and cost.

Fog computing is an emerging paradigm that brings the computing capabilities close to distributed IoT devices, which provides networking services between end devices and traditional cloud data centers. One important mission is to further reduce the monetary cost of fog resources while meeting the ever-growing demand of multiple users. In this paper, we focus on minimizing the total cost for multiple mobile users to provide an efficient resource provisioning scheme in fog computing. The total cost includes two aspects: the replication cost and the transmission cost. We consider two cases for the resource provision problem by focusing on different cost models. First, one simple case where users can only upload one replication is discussed, and an optimal solution is proposed by converting the original problem into one of bipartite graph matching. Then we consider a more complicated case that each user can upload multiple replications on fog nodes in the resource provisioning. For different transmission cost models, the transmission cost is related to the distance of each pair of fog nodes. This problem is proven to be NP-hard. We first propose a non-adaptive algorithm which is proved to be bounded by 2/3W+1/3OPT. Another 3+ε-approximation algorithm is proposed based on local search, which has better performance with higher complexity. Extensive simulations also prove the efficiency of our schemes.

Urban environmental monitoring related to such issues as air pollution and noise helps people understand their living environments and promotes urban construction. It is more and more important nowadays. By crowdsourcing, we can get mobile users at a low cost to collect measurement at different locations. This paper studies how to select optimal mobile users to construct an accurate monitoring map under a limited budget. We extend the noise Gaussian Process model to construct the data utility model. Because the monitoring map is updated in each time slot, we try to maximize the time-averaged data utility under the time-averaged budget constraint. This problem is particularly challenging given the unknown future information and the difficulty of solving the one-slot problem: maximizing a non-monotone sub-modular objective under the budget constraint. To address these challenges, we first make use of Lyapunov optimization to decompose the long-term optimization problem into a series of real-time problems which do not require a priori knowledge about the future information. We then propose a time-efficient online algorithm to solve the NP-hard one-slot problem. As long as the algorithm for the one-slot problem has a competitive ratio e, the time-averaged data utility of our online algorithm has a small gap compared with e times the optimal one. Evaluations based on the real air pollution data in Beijing [2] and real human trajectory data [1] show the efficiency of our approach.

The ride-hailing system has become popular around the world. The Service Providers (SPs) such as Uber and Didi dispatch passenger orders based on their location information. However, one concern from the public is whether the SPs could protect the location privacy of passengers. In this paper, we propose an order dispatch scheme that could preserve the location privacy of passengers based on their requirements. Our scheme uses cloaking regions in which the SPs cannot distinguish actual locations of passengers. The trade-off is the loss of matching performance or social welfare, i.e., the increase in the overall pick-up distance. We formulate the problem as maximizing the social welfare (or minimizing the overall pick-up distances) under privacy requirements of passengers. A bipartite-matching-based scheme is investigated, and we provide a theoretical bound on the matching performance under specific privacy requirements. Nevertheless, minimizing the overall pick-up distances does not consider the interest of each individual passenger. Passengers with low privacy requirements may be matched with drivers far from them. Therefore, we further propose a pricing scheme that could make up for the individual loss by allocating discounts on their riding fares. Especially, three discount allocation strategies are proposed in this paper. Experiments on both real-world and synthetic datasets show the efficiency of our scheme.

Mobile crowdsensing emerges as a powerful sensing paradigm which utilizes a group of users with mobile devices to perform sensing tasks cooperatively. The traditional mobile crowdsensing architecture is centralized and cloud-based, where the users upload the sensing data directly to the central server. Due to the fact that the amount of sensing data is very large, it will bring much burden to upload and process data collected by mobile users on the central server. In this paper, we propose an edge-based mobile crowdsensing architecture, which introduces a new intermediate layer for data storage, processing and aggregation through deploying mobile edge servers between the traditional cloud server and the user layer. Considering the limited budget, a decision-maker has to decide which server is activated to process each data type at minimum cost, where the cost consists of the facility cost for activating server and processing data, and the service cost measured by the distance of mobile users' movement during the process of uploading data. Since a user may have multiple types of data, this problem is formulated as a variant of the uncapacitated multi-commodity facility location problem. Furthermore, an approximation algorithm is proposed to solve it for minimizing cost, which is proved to have a bound to the optimal solution. We conduct extensive simulations based on the widely-used real-world datasets: roma taxi set, epfl mobility set and geolife trajectory set. The experiment results show that the proposed approximation algorithm outperforms other baseline algorithms and accords with the theoretical bound results.

With the progress of mobile devices and the successful usage of the wisdom of crowds, spatial crowdsourcing has attracted much attention from the research community. This paper addresses the efficient worker recruitment problem under the task coverage constraint. The efficiency of worker recruitment is measured by the total quality collected by a set of workers and the corresponding cost, e.g., proportional to the overall trajectory length of workers. Specifically, we consider two different scenarios, 1-D line topology and general 2-D topology, in which workers may have either homogeneous or heterogeneous crowdsourcing quality (e.g., the quality of videos or photos for an object at a particular location). In the 1-D scenario, we propose two dynamic programming approaches to find the optimal solution in both homogeneous and heterogeneous cases. In the general 2-D scenario, the proposed problem turns out to be NP-hard even in the homogeneous case. We first prove that the simple nearest assignment has an approximation ratio of 1/(2n), where n is the number of the workers. Therefore, the nearest assignment cannot be scalable. We further propose a novel assignment approach based on the minimum spanning tree. The proposed approach is proved to be close to the optimal solution in the homogeneous case and 1/ρ in the heterogeneous case, where ρ is the maximum quality ratio between two workers. The effectiveness of the proposed algorithm is verified using a real mobility trace: Uber pick-up trace in the New York City.

Recently, the development of Bike Sharing Systems (BSSs) brings environmental and economic benefits to the public. However, BSSs frequently suffer from the imbalanced bike distribution, including dock-less BSSs. The underflow or overflow of bikes in a region may lead to a lower service level to BSSs or congestion to the city. In the paper, we consider rebalancing the dock-less BSS by providing users with monetary incentives. The long-term objective is to maximize the number of satisfied users who successfully complete their rides over a period of time. The operator of the dock-less BSS can not only encourage a user to rent bikes at the neighborhood of its source with a source incentive, but also incentivize them to return bikes at the neighborhood of its destination with a destination incentive. To learn the differentiated incentive price for rebalancing bikes across time and space, we extend a novel deep reinforcement learning framework for user incentive. The source and destination incentives are integrated in an adaptive way by adjusting the detour level at the source and/or destination by avoiding bike underflow and overflow. In the experiment, we evaluate our approach in comparison with two existing pricing schemes. The locations of sources and destinations are abstracted from a selected dataset from Mobike. The experiment results show that our adapted learning algorithm outperforms the original one that only considers source incentive as well as another state-of-the-art approach in maximizing the long-term number of satisfied users.

Network function virtualization (NFV) enables flexible implementation of software middleboxes (network functions) as virtual machines running on standard servers. However, the flexibility also creates a challenge for efficiently placing such VNFs, due to the availability of multiple hosting servers and VNF capability of changing traffic volumes. In this paper, we address the placement problem of a special type of VNF with the traffic-diminishing effect (e.g., spam filters). We aim at minimizing the total bandwidth consumption of flows by placing at most a pre-determined number of VNFs. First, we formulate the Traffic-diminishing VNF Placement (TVP) as an optimization problem. We propose an optimal strategy for the tree-structured networks. Next, we extend the algorithm to the directed acyclic graph (DAG) topology. Then an approximation algorithm is designed after we prove the NP-hardness of our problem in a general topology. Extensive simulations are conducted on CAIDA dataset to evaluate the performance of our proposed algorithms in various scenarios.

Voice-based input is usually used as the primary input method for augmented reality (AR) headsets due to immersive AR experience and good recognition performance. However, recent researches have shown that an attacker can inject inaudible voice commands to the devices that lack voice verification. Even if we secure voice input with voice verification techniques, an attacker can easily steal the victim's voice using low-cast handy recorders and replay it to voice-based applications. To defend against voice-spoofing attacks, AR headsets should be able to determine whether the voice is from the person who is using the AR headsets. Existing voice-spoofing defense systems are designed for smartphone platforms. Due to the special locations of microphones and loudspeakers on AR headsets, existing solutions are hard to be implemented on AR headsets. To address this challenge, in this paper, we propose a voice-spoofing defense system for AR headsets by leveraging both the internal body propagation and the air propagation of human voices. Experimental results show that our system can successfully accept normal users with average accuracy of 97% and defend against two types of attacks with average accuracy of at least 98%.

Software middleboxes, also called Virtual Network Functions (VNFs), are replacing expensive traditional hardwares in implementing network services. Multiple middleboxes processing flows in a specific order form a service chain. Current works mostly focus on the service chain deployment problem and pay little attention to the flow scheduling of a deployed service chain, resulting in a poor control of the flow completion times. However, a high performance network always has a strict requirement of the flow completion time. In this paper, we build a transmission and processing delay model to formulate the communication latency behavior of flows being processed by middleboxes. We aim to minimize the flow completion time in two aspects: the longest completion time (makespan) and the average completion time. We propose an optimal solution for each aspect when there are only two middleboxes in the service chain. With a service chain with an arbitrary length, we first prove the NP-hardness of our problem in both aspects and then design two corresponding heuristic algorithms, which are extended from our proposed optimal solutions for a service chain with a length of two. Extensive simulations are conducted to evaluate the performance of our algorithms in various scenarios.

Motivated by many Online Social Network (OSN)applications such as viral marketing, the Social Influence Maximization Problem (SIMP)has received tremendous attention. SIMP aims to select k initially-influenced seed users to maximize the number of eventually-influenced users. Under the independent cascade model, the SIMP has been proved to be NP-hard, monotone, and submodular. Therefore, a naive greedy algorithm that maximizes the marginal gain obtains an approximation ratio of 1-e-1. This paper extends the SIMP by considering the crowd influence which is combined group influence in additional to individual influence among a given crowd. Our problem is proved to be NP-hard and monotone, but not submodular. It is proved to be inapproximable within a ratio of |V|ε-1for any ε > 0. However, since user connections in OSNs are not random, approximations can be obtained by leveraging the structural properties of OSNs. We prove that the supmodular degree, denoted as Δ. of most OSNs has the following property lim|V|→∞[Δ/O(|V|)] = 0, i.e., Δ ∈ o(|V|) for most OSNs. The supermodularity, denoted by \triangle, is used to measure to what degree our problem violates the submodularity. Two approximation algorithms have been applied with ratios of 1/(Δ+2) and 1-e-1/(Δ+1), respectively. Experiments demonstrate the efficiency and effectiveness of our algorithms.

Unattended wireless sensor networks (UWSNs) are vulnerable to mobile sink (MS) replication attack. In this attack, using the compromised key information, an attacker can collect data from networks by impersonating sinks. To resist such an attack, some schemes have been proposed. To improve the resilience of MS replication attack of these schemes, we can integrate them with schemes based on deployment knowledge. However, there are the following defects: 1) the probability of mutual authentication between a MS and a sensor node is less than 1 and 2) during the authentication phase, the energy consumption of sensor nodes increases significantly as the deployment area expands. In this paper, we construct 3-D backward key chains based on deployment knowledge and propose a new authentication scheme based on these. As compared with these existing related schemes, the detailed theory analysis and simulation results indicate that the scheme can ensure that a MS can be authenticated by sensor nodes, and can improve the resilience of networks' MS replication attack with low energy consumption.

Software-Defined Networks (SDNs) introduce great flexibilities in how packet routes can be defined and changed over time, and enable a more fine-grained and adaptive traffic engineering. The recently introduced support for more accurate synchronization in SDNs further improves the degree of control an operator can have over the packets' forwarding paths, and also allows to avoid disruptions and inconsistencies during network updates, i.e., during the rerouting of flows. However, how to optimally exploit such technology algorithmically - to efficiently schedule the update of multiple flows in such timed SDNs - while accounting for possible interference and congestion, is not well-understood today. We, in this paper, initiate the study of the fundamental problem of how to reroute the updates of multiple network flows in a synchronized SDN in a congestion-free manner. We rigorously prove that the problem is NP-hard for flows of unit size and network links with unit delay. We also show that a greedy approach to update the network can delay the update significantly. Our main contribution is the first solution to this problem: Chronicle. Our approach is based on time-extended network construction and the resource dependency graph, which is implemented by Openflow 1.5 using the scheduled bundles feature. The evaluation results show that Chronicle can reduce the makespan by 63% and reduce the number of changed rules by 50% compared to state-of-the-art.

The mission of set reconciliation (also called set synchronization) is to identify those elements which appear only in exactly one of two given sets. In this paper, we extend the set reconciliation problem into three design rationales: (i) multiset support; (ii) near 100 percent reconciliation accuracy; and (iii) communication-friendly and time-saving. These three rationales, if realized, will lead to unprecedented benefits for the set reconciliation paradigm. Generally, prior reconciliation methods are mainly designed for simple sets and thus remain inapplicable for multisets. Methods based on probabilistic data structures, e.g., the Counting Bloom Filter (CBF), support efficient representation, and multiplicity queries. Based on these probabilistic data structures, approximate multiset reconciliation can be enabled. However, they often cannot achieve a statisfying accuracy, due to potential hash collisions. The reconciliations enabled by logs or lists incur high time-complexity and communication overhead. Therefore, existing reconciliation methods, fail to realize the three rationales simultaneously. To this end, we redesign Trie and Fenwick Tree (FT), to near-accurately represent and reconcile two types of multisets that we refer to as unsorted and sorted multisets, respectively. Moreover, to further reduce the communication overhead during the reconciliation process, we design a partial transmission strategy when exchanging two Tries or FTs. Comprehensive evaluations are conducted to quantify the performance of our proposals. The trace-driven evaluations demonstrate that Trie and FT achieve near-accurate multiset reconciliation, with 4.31 and 2.96 times faster than the CBF-based method, respectively. The simulations based on synthetic datasets further indicate that our proposals outperform the CBF-based method in terms of accuracy and communication overhead at most time.

Due to instability of links in the network, the model of delaytolerant networks (DTNs) is often adopted in many emerging mobile applications. To organizations and individuals using these mobile applications, security and privacy are considered to be two of the most significant concerns. This research seeks to address anonymous communications in DTNs. While many different anonymous routing protocols have been proposed for ad hoc networks, to the best of our knowledge, only variants of onion-based routing have been tailored for DTNs. Since each type of anonymous routing protocol has its pros and cons, there is no single anonymous routing protocol for DTNs that can adapt to the different levels of security requirements. In this paper, we first design a set of anonymous routing protocols for DTNs, called anonymous Epidemic and zone-based anonymous routing, based on the original anonymous routing protocols for ad hoc networks. Then, we propose a framework of anonymous routing (FAR) for DTNs, which subsumes all the aforementioned protocols. By tuning its parameters, the proposed FAR is able to outperform onion-based, anonymous Epidemic, and zone-based routing. In addition, numerical analyses for the traceable rate, node anonymity, and path anonymity models are built. Extensive simulations using randomly generated graphs as well as real traces demonstrate that the proposed framework for DTNs successfully achieves its design goals.

With computer vision-based technologies, current Augmented reality (AR) systems can effectively recognize multiple objects with different visual characteristics. However, only limited degrees of distinctions can be offered among different objects with similar natural features, and inherent information about these objects cannot be effectively extracted. In this paper, we propose TaggedAR, i.e., an RFID-based approach to assist the recognition of multiple tagged objects in AR systems, by deploying additional RFID antennas to the COTS depth camera. By sufficiently exploring the correlations between the depth of field and the received RF-signal, we propose a rotate scanning-based scheme to distinguish multiple tagged objects in the stationary situation, and propose a continuous scanning-based scheme to distinguish multiple tagged human subjects in the mobile situation. By pairing the tags with the objects according to the correlations between the depth of field and RF-signals, we can accurately identify and distinguish multiple tagged objects to realize the vision of “tell me what I see” from the AR system. We have implemented a prototype system to evaluate the actual performance with case studies in a real-world environment. The experiment results show that our solution achieves an average match ratio of 91 percent in distinguishing up to dozens of tagged objects with a high deployment density.

Mobile CrowdSensing is a new paradigm in which requesters launch tasks to the mobile users who provide the sensing services. The tasks, in practice, are usually heterogeneous (have diverse spatial-temporal requirements), which make it hard to select an efficient subset of users to perform the tasks. In this paper, we present a point of interest (PoI) based mobility prediction model to obtain the probabilities that tasks would be completed by users. Based on it, we propose a greedy offline algorithm to select a set of users under a participant number constraint. Furthermore, we extend the user selection problem to a more realistic online setting where users come in real time and we decide to select or not immediately. We formulate the problem as a submodular k-secretaries problem and propose an online algorithm. Finally, we design a distributed user selection framework Crowd UserS and implement an Android prototype system as proof of the concept. Extensive simulations have been conducted on three real-life mobile traces and the results prove the efficiency of our proposed framework.

Wireless energy transfer has been widely studied in recent decades, with existing works mainly focused on maximizing network lifetime, optimizing charging efficiency, and optimizing charging quality. All these works use a charging model with the linear superposition, which may not be the most accurate. We apply a nonlinear superposition model, and we consider the Fast Charging Scheduling problem (FCS): Given multiple chargers and a group of sensors, how can the chargers be optimally scheduled over the time dimension so that the total charging time is minimized and each sensor has at least energy E ? We prove that FCS is NP-complete and propose a 2-approximation algorithm to solve it in one-dimensional (1D) line. In a 2D plane, we first consider a special case of FCS, where the initial phases of all chargers are the same, and propose an algorithm to solve it, which has a bound. Then we propose an algorithm to solve FCS in a general 2D plane. Unlike other algorithms, our algorithm does not need to calculate the combined energy of every possible combination of chargers in advance, which greatly reduces the complexity. Extensive simulations demonstrate that the performance of our algorithm performs almost as good as the optimal algorithm.

Edge computing is one of the emerging technologies aiming to enable timely computation at the network edge. With virtualization technologies, the role of the traditional edge providers is separated into two: edge infrastructure providers (EIPs), who manage the physical edge infrastructure, and edge service providers (ESPs), who purchase slices of physical resources (e.g., CPU, bandwidth, memory space, disk storage) from EIPs and then cache service entities to offer their own value-added services to end users. When an ESP caches a service entity in an edge server, the ESP has to pay some fees (i.e, the cache cost) to the EIP that owns the edge server. One of the fundamental problems in edge virtualization is the so-called service entity caching problem, i.e., where to place service entities for an ESP to minimize the cache cost. In this paper, we study the service entity caching problem from the utility perspective. We use `utility' to denote the positive impact on a client from caching a service entity in an edge server, and the exact meaning of utility can vary depending on specific scenarios. We formulate the Utility-based Service Entity Caching (UtilitySEC) problem, which can be generalized to many existing problems by modifying the `utility'. We prove that the UtilitySEC problem is NP-complete and design an approximation algorithm for it. Extensive simulations are conducted to evaluate the performance of the proposed framework.

The SDN control plane needs to frequently update the data plane as the network conditions change. Since each switch updates its flow table independently and asynchronously, the transition of data plane state - if done directly from the initial to the final stage - may result in serious flash congestion. Prior work strives to find a congestion-free update plan with multiple stages, each with the property that there will be no congestion independent of the update order. Yet congestion-free update may prevent the network from being fully utilized. It also requires solving a series of LP which is time-consuming. In this paper, we propose congestion-minimizing update and focus on two general problems: The first is to find routing at each intermediate stage that minimizes transient congestion for a given number of intermediate stages. The second is to find the minimum number of intermediate stages and an update plan for a given maximum level of transient congestion. We formulate them as two optimization programs and prove their hardness. We propose a set of algorithms to find the update plan in a scalable manner. Extensive experiments with Mininet show that our solution reduces update time by 50 percent and saves control overhead by 38 percent compared to prior work.

Due to software implementation failure and misuse of cryptography, data encryption can no longer be considered a safeguard from security attacks. As a result, adversaries with eavesdropping capability along a routing path can compromise data privacy. In addition, should an adversary be one of the intermediate relay nodes in a path, she can deny data forwarding to disconnect the end-to-end communications. One solution is to avoid message routing through certain insecure areas, such as malicious countries or likely-compromised nodes. To this end, an avoidance routing based on the single path has been proposed. However, this single-path-based protocol relies on the availability of a safe path, i.e., no adversary is in the proximity of the whole path, which is difficult to achieve and therefore limits the routing opportunity. To tackle this issue, we propose an avoidance routing framework, namely timer-based multi-path avoidance routing (TMPAR). In our approach, a source node first encodes a message into k different pieces, and each piece is sent via a different path. During its path discovery phase, a timer is used to efficiently discover a better set of paths. The destination can assemble the original message easily. Under the condition that no adversary obtains all the k pieces of the message, the proposed TMPAR can securely deliver a message to its destination in spite of eavesdropping. The extensive ns-2 simulation results demonstrate that our TMPAR achieves its design goals.

Edge computing is gaining popularity these years, more service providers are shifting their services from clouds to the edge for better QoS provision. Recent studies in NFV also tend to deploy Network Function Virtualization (NFV) services in the edge network. However, the NFV deployment in the edge network is a challenging problem and differs from the similar problem in data centers. We mainly focus on a new NFV Chain Placement (NCP) problem in the paper. It is well known that the edge of the network is dynamic, and edge computing aims to utilize the physical edge resources efficiently and quickly. We first prove that the NCP problem is NP-complete. Then we propose a novel metric that can better measure the balance condition of the physical resources. We also analyze its advantages. Finally, we design an efficient algorithm, MINI, based on this metric. We evaluate MINI using extensive simulations. The results show that MINI has great advantages over a genetic algorithm (GA) in terms of physical resource utilization, acceptance rate, and running time.

With the bloom of the sharing economy, bike sharing systems have earned increasing attention, and a great amount of bike sharing systems have been established in major cities. Users of these systems mainly conduct one-way trips, which leads to an unbalanced distribution of the bikes over time and space. The system operators could hire a fleet of vehicles to move bikes among bike stations for rebalancing. We focus on a routing schedule problem for each vehicle used in the rebalancing process and aims to minimize its moving distance. For the problem, we propose a greedy algorithm which can be easily extended to a parallel version. The scheduled route for a vehicle is adapted from the Hamiltonian path covering all unbalanced bike stations. The algorithm greedily adjusts the route if the vehicle cannot moving along the Hamiltonian path due to capacity limitation violation. Different from previous approaches, our algorithm has a more flexible tradeoff between running time consumption and optimality of the output. Finally, we conduct experiments on both real-world and synthetic datasets and compare the performance of our algorithm with a classic approach.

Software-defined networks (SDNs) can support finegrained forwarding policies in the underlying switches. The new content addressable memory, Ternary Content Addressable Memory (TCAM), enables fast lookups for matching rules in message forwarding, represented as binary strings with wildcards. However, the cost and power limit the number of matching rules a TCAM can support. Therefore, rule caching is needed to place high-weight (high-hit) rules in the TCAM hardware, while large, but slow, software switches handle cache-miss traffic. We assume that matching these rules form a forest of trees. A rule R' is a descendant of another rule R if R' is a special case of R. Dependent rules are evaluated in a particular matching order: when a rule is included in the cache, all its descendants in the rule set have to be included as well. Our objective is to maximize the number of rule hits, while limiting the number of cached rules. Three greedy rule-caching algorithms are proposed, including two with approximation ratios of 2 and24/5 , respectively. In addition, we propose a dynamic programming solution that is optimal but slow. The efficiency of the proposed approaches are evaluated through real data-driven simulations.

In data center networks (DCN), large scale flows produced by parallel computing frameworks form many coflows semantically. Most inter-coflow schedulers only focus on the remaining data of coflows and attempt to mimic Shortest Job First (SJF). However, a coflow may consist of multiple stages. In this paper, we consider the Multi-stage Inter-Coflow Scheduling problem and try to give an efficient online scheduling scheme. We first explore a short-sighted algorithm with the greedy strategy. This gives us an insight into utilizing the network resources. Based on that, we propose a far-sighted heuristic, which schedules sub-coflows to occupy network bandwidth in turn. Through simulations in various network environments, we show that, compared to a state-of-the-art scheduler - Varys, a multi-stage aware scheduler can reduce the coflow completion time by up to \pmb4.81× even though it is short-sighted. Moreover, the far-sighted scheduler can improve the performance by nearly \pmb7.95 × reduction.

Communication in data centers often involves many parallel flows that all share the same performance goal. A useful abstraction, coflow, is proposed to express the communication requirements of prevalent data parallel paradigms. The multiple coflow routing and scheduling problem faces challenges when deriving a good theoretical performance ratio because coexisting coflows will compete for the same network resources such as link bandwidths. In this paper, we focus on the coflow problem in the most popular data center infrastructure: the Leaf-Spine topology. We first formulate the problem and study the path selection issue on this two-tier structure. In order to minimize the average coflow completion time (CCT), we propose a Multi- hop Coflow Routing and Scheduling strategy (MCRS) for inter-coflows and intra-coflows and prove that our method has a reasonably good competitive ratio. Extensive experiments show that MCRS outperforms other state-of- art schemes.

Due to the fast increase of mobile traffic, most mobile network operators face the congestion issue in licensed spectrum bands. Several telecommunication vendors and operators propose to expand LTE service to the unlicensed spectrum bands to relieve the traffic congestion. However, LTE in unlicensed spectrum may interfere with Wi-Fi communications in the same bands and cause significant decrease in the quality of service of Wi-Fi. In this paper, we propose a novel mechanism that enables negotiations between two different wireless technologies (Wi-Fi and LTE), which ensures fair spectrum sharing between Wi-Fi and LTE-Unlicensed (LTE-U) in the same bands. We formulate the co-existence of Wi-Fi and LTE-U as a constrained optimization problem, and we solve the problem. We evaluate the performance of the proposed scheme via NS-3 simulations. The simulation results show that our approach can effectively improve the overall channel utilization and reduce the interference between Wi-Fi and LTE-U.

The rapidly increasing number of vehicles in roads leads to numerous problems in metropolitan areas. Several researchers show that carpooling can be an efficient solution to relieve the pressures caused by large numbers of cars. Previous research on carpools introduces several additional constraints to simplify the problem, but some of them are unreasonable in reality. In this paper, we focus on removing the static capacity constraint. Doing so allows a vehicle to carry more passengers than vehicle's capacity, which is possible if some people are dropped off and new passengers take their places during the journey. A greedy approach based on multi-round matching is proposed, and it is further improved by taking advantage of geometry properties. We apply our algorithms to both simulated and real world datasets, and experiment results show that our algorithms have better performances than existing approaches.

In cellular traffic offloading through opportunistic mobile networks, existing schemes rely on the assumption that data can be entirely transmitted at each contact. However, transmission probability exponentially decreases as data size increases. That is, the contact duration in each contact might be insufficient for delivering large data. The objective of this paper is to find an optimal traffic offloading scheme through data partitioning so that the data delivery latency is minimized. There is a trade- off in data partitioning. Each small chunk in a path has a higher delivery probability than original data, and consequently, has a shorter delivery latency under the persistent transmission model with re-transmission. However, the destination needs to receive all the chunks in multiple paths to retrieve the data. A delay in any path will lead to a longer delivery latency. We formulate the optimal cellular traffic offloading problem and propose an approach to generate forwarding paths with possible heterogeneous data chunks. Specifically, we discuss the optimal solution for single-hop direct forwarding with multiple offloading helpers and optimal chunk sizes. Then, we propose using the node's social-feature to generate multiple edge-disjoint multi-hop forwarding paths. Extensive experiments on realistic traces show that our scheme achieves a much better performance than those without partitioning.

Task resource allocation has always been an important issue in cloud-based data center networks (DCNs). This paper considers provisioning the maximum admissible load (MAL) of virtual machines (VMs) in physical machines (PMs) with underlying tree-structured DCNs using the hose model for communication. The limitation of static load distribution is that it assigns tasks to nodes in a once-and-for-all manner, and thus, requires a priori knowledge of program behavior. To avoid load redistribution during a run time where the load grows, we introduce maximum elasticity scheduling, which has the maximum growth potential subject to the node and link capacities. Given a tree-based topology, this paper aims to find the schedule with the maximum elasticity across both nodes and links. We have found a distributed linear solution based on message passing, and we discuss several extensions of the model. We conclude the paper by presenting various simulation results.

Mobile crowdsensing recruits a massive group of mobile workers to cooperatively finish a sensing task through their smart devices (mobile phones, ipads, etc.). In this paper, the communication in social network for delivering the sensing data of mobile crowdsensing is considered, where some requesters publish the sensing tasks to all the Point of Interests (PoIs), and the workers are recruited to take the sensing data in the PoI until they could communicate with the requester through an offline and online connection. We first use the semi-Markov model to predict the offline encounter situation. Then, the worker's utility is decided by both the offline encounter and social connection probabilities. The Worker Recruitment for Self-organized MSC (WEO) is further presented through recruiting a set of workers, who have the maximum communication probability with the requesters. We prove that the optimal recruitment problem is NP-hard, and we introduce a practical greedy heuristic method for this problem, the performance of the greedy method is also tested. Two real-world traces, roma/taxi and epfl are tested in our simulations, where WEO always achieves the highest delivery ratio of sensing tasks among different recruitment strategies.

Network Function Virtualization (NFV) evolves the implementation of network functions from expensive hardwares to software middleboxes. These software middleboxes, also called Virtual Network Functions (VNFs), are executed on switch-connected servers. Efficiently deploying such VNFs is challenging, because VNFs must fully process all flows with their traffic rates before they reach their destinations while VNF locations are restricted by the constraint of vertex capacity. In addition, each network function offers heterogeneous VNF types with different configurations of processing volumes and costs. This paper focuses on minimizing the total cost of deploying VNF instances for providing a specific network function to all flows in tree-structured networks. First we prove the NP-hardness of heterogeneous VNF deployment in a tree topology and propose a dynamic programming based solution with a pseudo-polynomial time complexity. Then we narrow down to three simplified cases by focusing on homogeneous VNFs or the linear line topology. Specifically, three algorithms are introduced: an improved dynamic programming based algorithm for deploying homogeneous VNFs in a tree topology, a performance-guaranteed algorithm for deploying heterogeneous VNFs in a linear line topology, and an optimal greedy algorithm for deploying homogeneous VNFs in a linear line topology. Extensive simulations are conducted to evaluate the performance of our algorithms.

A denial-of-service (DoS) attack is a cyber-attack in which the attacker sends out a huge number of requests to exhaust the capacity of a server, so that it can no longer serve incoming requests and DoS occurs. The most devastating distributed DoS attack is performed by malicious programs called bots. With the help of a special type of router called filter router, the victim can protect itself and reduce useless congestion in the network. A server can send out filters to filter routers for blocking attack traffic. The victim needs to select a subset of filter routers wisely to minimize attack traffic and blockage of legitimate users (LUs). In this paper, we formulate two problems for selecting filter routers given a constraint on the number of filters. The first problem considers the source-based filter and we provide greedy approximation solutions. The second problem considers the destination-based filter and how to minimize total amount of attack traffic and blocked LUs. We propose a dynamic programming solution for the second problem. We present simulation results comparing the proposed solutions with a naive approach. Our simulation results strengthen support for our solutions.

Recently, fog computing has become a highly virtualized platform that provides computation, storage, and networking services between end devices and traditional cloud data centers. In this paper, we address the resource provision (RP)problem for delay-sensitive users in cooperative fog computing. Our objective is to find a feasible provision scheme that minimizes the total monetary cost proportional to the number of fog nodes for network operators under the deadline and capacity constraints by considering the cooperation of fog nodes. We consider two cases of our RP problem: the Unlimited-Processor Fog Nodes (UPFN)case and the Limited-Processor Fog Nodes (LPFN)case. For the UPFN case, each fog node has unlimited processors. The requests on each fog node can be processed in parallel ideally, i.e. with no scheduling delay. The LPFN case corresponds to a more realistic scenario where the scheduling delay is non-eligible. In either case, our RP problem is proven to be NP-hard. For the UPFN case, we propose two greedy algorithms which iteratively remove fog nodes according to their global or local cooperative influences until there is no feasible provision that can guarantee users' deadlines. For the LPFN case, it is not trivial to check the existence of a feasible provision due to the interactive influence on the scheduling delay for requests. We find a near-optimal solution with bound [8/3]OPT+[(ε2)/(8mα)] using the continuous congestion game and check the feasibility, where m is the number of fog nodes and α is a constant value related to the delay function. Extensive simulations demonstrate the efficiency of our schemes.

Voiceprint-based authentication is fast becoming the everyday norm since it is much easier to use and provides better security. However, current voiceprint-based authentication systems are vulnerable to various replay attacks. To tackle the spoofing attacks, we propose a new system that leverages the structural differences between human vocal system and loudspeakers and use the unique vibration pattern of both human vocal cord and throat as a key differentiating factor for liveness detection. Specially, we model the relationship between voices collected by two microphones of a smartphone of each live speaker using sparse representation. Compared with existing systems, our solution does not assume any prior knowledge of the attack method and is easy to operate. Moreover, our solution leverages the audio signals within the vocal frequency range and is robust to jamming attacks using high-frequency audio. Experimental results show that our system can achieve accurate live ness detection for a 6-digit passphrase with a mean true acceptance rate of 99.04% and true rejection rate of 100%.

Network Function Virtualization (NFV) changes the way that we implement network services, or middleboxes, from expensive hardwares to software functions. These software middleboxes, also called Virtual Network Functions (VNFs), run on switch-connected commodity servers. Efficiently placing such middleboxes is challenging because of their traffic-changing effects and dependency relations. Private (used by one single flow) middleboxes can save more link bandwidth resources while shared (used by multiple flows) middleboxes cut down server resource expenses. This paper formulates the resource usage trade-off between bandwidth consumption and cost of middlebox placement as a combined cost minimization problem. After proving the NP-hardness of our problem in general topologies, we narrow down to a specific kind of topology: tree-structured networks. We study two kinds of constraints: traffic-chaining ratio and middlebox dependency relations. With homogeneous flows, we propose optimal greedy algorithms for the placement of a single middlebox first, and then multiple middleboxes without order. We also introduce a dynamic programming algorithm for the placement of a totally-ordered middlebox set. A performance-guaranteed algorithm is designed to handle heterogeneous flows. Extensive simulations are conducted to evaluate the performance of our proposed algorithms in various scenarios.

Fair allocation has been studied intensively in both economics and computer science. Many existing mechanisms that consider fairness of resource allocation focus on a single resource. With the advance of cloud computing that centralizes multiple types of resources under one shared platform, multi-resource allocation has come into the spotlight. In fact, fair/efficient multi-resource allocation has become a fundamental problem in any shared computer system. The widely-used solution is to partition resources into bundles that contain fixed amounts of different resources, so that multiple resources are abstracted as a single resource. However, this abstraction cannot satisfy different demands from heterogeneous users, especially on ensuring fairness among users competing for resources with different capacity limits. A promising approach to this problem is dominant resource fairness (DRF), which tries to equalize each user's dominant share (share of a user's most highly demanded resource, i.e., the largest fraction of any resource that the user has required for a task), but this method may still suffer from significant loss of efficiency (i.e., some resources are underused). This paper develops a new allocation mechanism based on DRF aiming to balance fairness and efficiency. We consider fairness not only in terms of a user's dominant resource, but also in another resource dimension which is secondarily desired by this user. We call this allocation mechanism 2-dominant resource fairness (2-DF). Then, we design a non-trivial on-line algorithm to find a 2-DF allocation and extend this concept to k-dominant resource fairness ( k-DF).