Gradient perturbation mechanisms, such as differential privacy (DP), aim to defend against gradient inversion attacks (GIA) by injecting noise into the shared gradients. Recent studies have shown that DP-based defenses lack robustness against advanced GIAs. However, existing gradient inversion methods typically rely on iterative refinement and assume static noise, resulting in low efficiency and limited reconstruction fidelity under high-noise conditions. In this paper, we propose Venom, a novel gradient inversion attack method based on a liquid diffusion mechanism. Venom reconstructs private data directly from DP-protected gradients without requiring any prior knowledge of the noise distribution. Specifically, we design a Structural Prior Extraction (SPE) module that analytically extracts deep feature representations from perturbed gradients through energy-based aggregation, enabling stable pre-reconstruction of users' latent data features. We further introduce a Diffusion-driven Liquid Recovery Network (Diff-LRN) for high-fidelity image reconstruction. Unlike traditional diffusion models that rely on iterative sampling with predefined noise schedules, Diff-LRN performs deterministic single-step reconstruction using adaptive liquid neural dynamics to handle spatially heterogeneous noise patterns. Experiments across four benchmarks demonstrate that Venom achieves a speedup of up to 38,343× over state-of-the-art attacks while maintaining high reconstruction fidelity under strong DP settings. These results challenge prevailing assumptions about DP robustness and underscore the need for more resilient privacy-preserving mechanisms in federated learning.

Data-protection regulations such as the GDPR grant every participant in a federated system a right to be forgotten. Federated unlearning has therefore emerged as a research frontier, aiming to remove a specific party's contribution from the learned model while preserving the utility of the remaining parties. However, most unlearning techniques focus on Horizontal Federated Learning (HFL), where data are partitioned by samples. In contrast, Vertical Federated Learning (VFL) allows organizations that possess complementary feature spaces to train a joint model without sharing raw data. The resulting feature-partitioned architecture renders HFL-oriented unlearning methods ineffective. In this paper, we propose ReMisVFU, a plug-and-play representation-misdirection framework that enables fast, client-level unlearning in splitVFL systems. When a deletion request arrives, the forgetting party collapses its encoder output to a randomly sampled anchor on the unit sphere, severing the statistical link between its features and the global model. To maintain utility for the remaining parties, the server jointly optimizes a retention loss and a forgetting loss, aligning their gradients via orthogonal projection to eliminate destructive interference. Evaluations on public benchmarks show that ReMisVFU suppresses back-door attack success to the natural class-prior level and sacrifices only about 2.5% points of clean accuracy, outperforming state-of-the-art baselines.

Vertical multimodal federated learning (VMFL) enables multiple clients holding data from different modalities to conduct collaboratively model training. Existing methods typically assume that multimodal data samples (i.e., text and image) from the same entity (i.e., person) are paired across the clients (i.e., aligned). However, this assumption rarely holds in practice, as data is often collected independently with no shared identifiers. To address this challenge, we propose hashing-based alignment (HAL), a new VMFL framework that works without pre-aligned samples. HAL consists of two key components. The first component is an efficient and privacy-preserving method to identify similar samples from different modalities as aligned pairs. It adopts locality sensitive hashing (LSH) for the efficient retrieval of similar samples, introduces a shift-orthogonal hashing scheme to tackle the gaps between different modalities, and uses a bloom-style method for secure Hamming distance estimation. We prove that the shift-orthogonal hashing reduces distance estimation errors and secure Hamming distance estimation satisfies differential privacy. The second component is a neighbor-aware fusion strategy, which applies cross-attention to aggregate informative signals from the aligned samples without relying on explicit similarity scores. Experimental results on two real-world datasets show that compared with five state-of-the-art (SOTA) baselines, HAL improves the cross-modal retrieval accuracy by over 63%, while also achieving up to 154× speedup.

Deep Neural Networks (DNNs) are widely used in intelligent applications, driving increasing computational demands on GPUs. However, modern GPU multitasking scheduling algorithms fail to effectively balance real-time task performance and resource utilization, especially under dynamic workloads with highly variable DNN computational demands. The complex and workload-dependent execution times of DNN kernels often lead to inefficient resource allocation, degraded system throughput, and missed real-time constraints. To address these challenges, we propose Flexible Parallel Orchestrator (FlePo), a GPU multitasking scheduling framework designed to optimize resource utilization and maintain real-time task performance within acceptable limits for soft real-time systems. FlePo integrates two key techniques: Adaptive Padding Dispatch (APD), which dynamically schedules best-effort tasks while leveraging the predictable execution characteristics of DNN kernels to maintain real-time predictability; and Dynamic Parallel Fusion (DPF), which employs kernel fusion to create computational isolation, reducing interference in parallel job execution. By combining offline profiling with online adaptation, FlePo efficiently responds to workload variations. We evaluate FlePo on two heterogeneous GPU platforms, NVIDIA Tesla V100 and AMD MI50, achieving up to a 50% increase in throughput while keeping real-time overhead below 2%. Our work enhances GPU multitasking in dynamic environments, with potential applications in autonomous driving, smart homes, and intelligent healthcare.

Keeping runtime alive is critical for mitigating cold start issues in Function-as-a-Service (FaaS) platforms. State-of-the-Art (SOTA) keep-alive policies often draw an analogy to data caching, adapting classic replacement algorithms to manage runtime pools. Our analysis reveals that this analogy is fundamentally flawed: 1) they violate their own eviction priorities due to conflicts with running containers, and 2) they ignore the exorbitant memory-time resource cost of runtime replacement. To address these challenges, we present FaaShadow, a lightweight keep-alive policy that recasts the problem from simple caching to cost-aware resource allocation. FaaShadow introduces the concept of a shadow pool, a per-function data structure that enables online estimation of the marginal utility of memory adjustments. By quantifying both the potential performance gain from allocating new containers and the performance loss from removing existing ones, FaaShadow makes data-driven reallocation decisions that maximize the global warm start rate. Experimental results show that FaaShadow achieves a 95% warm start rate using only 60% of the memory required by the best-in-class baseline. When paired with our dynamic scaling mechanism, FaaShadow reduces average memory consumption by a staggering 81.71% compared to the SOTA predictive scaler, while upholding performance targets.

Short video platforms face significant challenges in efficiently delivering diverse video analytics to a concurrent user base. These analytics impose substantial storage and computation demands on edge devices, while offloading tasks to the cloud incurs high data transmission overhead. Edge-cloud collaborative computing, which distributes workloads between edge devices and the cloud, offers a promising solution. However, applying this paradigm to short video platforms encounters efficiency and scalability hurdles due to the limited memory capacity of commercial smartphones, the large volume of intermediate analytics data, and the highly dynamic workloads driven by extensive user activity. This paper introduces FaaSVA, a Function-as-a-Service (FaaS)-based edge-cloud video analytics system for short video platforms. FaaSVA enables efficient and scalable video analytics by integrating edge-cloud collaboration with serverless computing. On the edge, it implements a re-utilizing memory provisioning mechanism to allocate, release and reclaim memory for video frames, reducing intermediate data and minimizing the memory footprint on smartphones. On the cloud, a serverless framework supports rapid and elastic adaptation to fluctuating workloads. The system abstracts video analytics functions and utilizes a dynamic, workload-aware Directed Acyclic Graph (DAG) partitioning algorithm to intelligently schedule functions to optimal execution nodes, balancing computation and communication. We implement FaaSVA and show that FaaSVA can reduce video analytics end-to-end latency by 28.46%.

To tackle the challenge of data heterogeneity in federated learning (FL), personalized FL has been proposed to maximize individual utility (model performance) by customizing personalized models for clients. Considering the significance ofindividual rationality, existing works have formulated clients' participation decisions problem ashedonicgames. However, they assume that clients can participate in only one collaborative coalition, constraining players' attempts to join multiple coalitions. Different from prior works, we approach personalized FL from the perspective of hedonicoverlapping coalition formation(OCF) games where rational clients can join multiple coalitions and generate their personalized model by weighting the local and coalition models. Nevertheless, the key challenge in analyzing the game is how to achieve a stable coalition structure where no clients would deviate from the current structure. This leads to our main question:what does a stable OCF structure look like?To address this problem, we first investigate the linear FL models for theoretical insights. Then, we design a heuristic algorithm for achieving anindividually stableOCF structure. Experimental results demonstrate the feasibility of our algorithm for both linear and non-linear models, and show that our mechanism can improve the personalized model performance by up to 19% over existing methods.

Mobile applications such as browsers, video, and other interactive software are tightly coupled with frame rendering, which is critical for user experience. Frame rendering requires the collaboration of CPU, GPU, and memory to ensure energy efficiency and maintain Quality of Experience (QoE). However, there are notable deficiencies in resource management for these components. Our observations reveal three critical issues: 1) the operating system fails to accurately identify rendering-related threads, 2) thread groups lack strict resource constraints, leading to insufficient resources for rendering-related threads, and 3) frequency scaling across components is not coordinated, resulting in performance degradation and power inefficiencies. To address these issues, we propose Trident, a holistic resource management framework. Trident includes a cross-layer thread tracer to identify rendering-related threads, a reinforcement learning-based governor to coordinate the frequency of multiple hardware components, and a gain scheduling-based share controller to constrain resources among thread groups dynamically. Our framework aims to minimize power consumption while maintaining QoE. We implement Trident as a system service on five distinct smartphones, from older models to recent flagships, and evaluate its effectiveness on popular applications under various workloads. The results demonstrate that Trident reduces power consumption by up to 16.8% compared to three state-of-the-art techniques while ensuring QoE on mobile platforms. Additionally, the overhead introduced by Trident is minimal, making it an efficient solution for real-world deployment.

With the development of mobile and edge computing, the demand for low-bit quantized models on edge devices is increasing to achieve efficient deployment. To enhance the performance, it is often necessary to retrain the quantized models using edge data. However, due to privacy concerns, certain sensitive data can only be processed on edge devices. Therefore, employing Quantization-Aware Training (QAT) on edge devices has become an effective solution. Nevertheless, traditional QAT relies on the complete dataset for training, which incurs a huge computational cost. Coreset selection techniques can mitigate this issue by training on the most representative subsets. However, existing methods struggle to eliminate quantization errors in the model when using small-scale datasets (e.g., only 10% of the data), leading to significant performance degradation. To address these issues, we propose QuaRC, a QAT framework on edge devices via Relative entropy coreset selection and Cascaded layer correction. The framework consists of two main phases: In the coreset selection phase, QuaRC introduces the “Relative Entropy Score” to identify the subsets that most effectively capture the model's quantization errors. During the training phase, QuaRC employs the Cascaded Layer Correction strategy to align the intermediate layer outputs of the quantized model with those of the full-precision model, thereby effectively reducing the quantization errors in the intermediate layers. Experimental results demonstrate the effectiveness of our approach. For instance, when quantizing ResNet-18 to 2-bit using a 1% data subset, QuaRC achieves a 5.72% improvement in Top-1 accuracy on the ImageNet-1 K dataset compared to state-of-the-art techniques.

In real-time edge-cloud video analytics systems, the edge conducts initial analytics on the video frames to a split layer of a trained neural network model. Then, it sends intermediate results to the cloud for follow-up analytics. In this article, we first show that malicious attackers can perform reconstruction attacks and attribute inference attacks on those intermediate results. We present Preva, a new P rivacy-preserving R eal-time E dge-cloud V ideo A nalytics system that defends against both attacks while respecting latency constraints. Preva first applies a lightweight, policy-based video frames transformation scheme generated on the fly by PrevaNetv2, with a lightweight backbone and an early-exit mechanism that adapts the computation resources of users. To guarantee end-to-end latency, a contextual multi-armed-bandit algorithm (BBSplit) dynamically chooses the optimal split layer, and a frame-similarity filter (PPReuse) amortizes policy generation across video frames. We present a formal privacy analysis and show that Preva can guarantee privacy leakage under reconstruction attacks and attribute inference attacks. We evaluate Preva through three video analytics applications and show that Preva outperforms existing systems by 41.6% in analytics accuracy and 52.7% in privacy leakage.

Vertical federated learning (VFL) trains model when the features of data samples are scattered over multiple clients. To improve efficiency, a promising approach is to find a coreset of the data samples and use it as a smaller training set. However, existing methods produce a large coreset when there are many clients and have long running time. To address these problems, we propose HaCore for efficient coreset construction in VFL setting. HaCore first employs locality sensitive hashing (LSH) to map features to bit signatures locally on the clients, and then merges the local signatures for k-medoids clustering. Data samples that correspond to the medoids are added to the coreset. The core idea is that the distance of original data samples can be approximated by the Hamming distance between their LSH-based bit signatures. To accelerate k-medoids, we utilize an inverted index to search the nearest medoid and a bit-counting method to quickly compute the aggregate distance from many signatures to a medoid. We evaluate HaCore on 5 datasets and compare with state-of-the-art coreset construction methods for VFL. The results show that HaCore accelerates the best-performing baseline by over 45x and matches the accuracy of training with all samples.

Attribute Inference Attacks (AIAs) pose a significant threat to recommendation systems (RS) by enabling adversaries to use threat models to infer sensitive user attributes like gender or race from user embeddings, resulting in privacy breaches such as unauthorized profiling and discriminatory policies against specific groups. Existing attribute protection methods are primarily applied during training, suffering from significant limitations, such as architectural inflexibility, dependence on interaction data, and potential catastrophic degradation in recommendation performance. To overcome these challenges, we propose AttrCloak, an efficient and effective post-training attribute unlearning (AU) framework that removes sensitive information from user embeddings without altering RS training architectures. AttrCloak employs dual-objective optimization with parameter self-sharing to minimize mutual information between user embeddings and sensitive attributes while preserving recommendation quality. Furthermore, it accommodates data-free scenarios by leveraging regularization loss when interaction data is unavailable. Comprehensive evaluations on four real-world datasets demonstrate AttrCloak's good performance in privacy protection and recommendation performance.

Distributed tracing is widely employed for troubleshooting distributed systems such as microservices. Existing tracing systems typically improve one or more of data completeness, non-intrusiveness, or lightweight operation through various data generation strategies. However, no current solution optimizes all three aspects simultaneously, which can introduce significant overhead in I/O-intensive environments that demand both high completeness and minimal intrusion. In this paper, we introduce LightTrace, a novel, eBPF-enabled toolkit that optimizes the data transmission mechanism in distributed tracing. LightTrace integrates seamlessly with mainstream tracing systems and leverages eBPF to reduce end-to-end latency and lower overall system overhead. LightTrace is implemented using a combination of kernel-level eBPF and user-space Golang components. Our evaluation demonstrates that LightTrace decreases the average latency overhead by up to 22.8 % and improves the peak throughput of microservice systems by between$\mathbf{1 2. 3 \%}$and$\mathbf{1 8. 6 \%}$. Furthermore, LightTrace's adaptability across diverse tracing platforms underscores its versatility in various microservice environments.

Text-attributed graph (TAG) provides a text description for each graph node, and few- and zero-shot node classification on TAGs have many applications in fields such as academia and social networks. Existing work utilizes various graph-based augmentation techniques to train the node and text embeddings, while text-based augmentations are largely unexplored. In this paper, we propose Text Semantics Augmentation (TSA) to improve accuracy by introducing more text semantic supervision signals. Specifically, we design two augmentation techniques, i.e., positive semantics matching and negative semantics contrast, to provide more reference texts for each graph node or text description. Positive semantic matching retrieves texts with similar embeddings to match with a graph node. Negative semantic contrast adds a negative prompt to construct a text description with the opposite semantics, which is contrasted with the original node and text. We evaluate TSA on 5 datasets and compare with 13 state-of-the-art baselines. The results show that TSA consistently outperforms all baselines, and its accuracy improvements over the best-performing baseline are usually over 5%. The code is at https://github.com/wyx11112/TSA.

Federated Unlearning (FU) addresses the "right to be forgotten" in federated learning by removing specific client data's contribution without retraining from scratch. Existing FUs are data-dependent, which make the assumption that systems can access original training data or stored historical parameter updates during unlearning. However, the assumption cannot always hold in practice, as users usually request the deletion of client data and historical parameter updates due to privacy concerns or storage limitations. Therefore, it is crucial to develop a zero-shot FU method without such data access. The key challenge is how to distinguish and remove the impact of target clients without data-level information. Motivated by the idea that if we can learn client-specific personalized information from the model instead of data, FU can be model-centric and data-free, we present the first zero-shot FU framework ZeroFU. By embedding client contributions into the model during learning via condition computation, ZeroFU enables the model to possess personalized features for unlearning. The unlearning is achieved using a proposed GAN-based distillation framework that obfuscates the personalized feature of the target client. Evaluations demonstrate its effectiveness in unlearning under non-IID settings.

Split learning is a prominent framework for vertical federated learning, where multiple clients collaborate with a central server for model training by exchanging intermediate embeddings. Recently, it is shown that an adversarial server can exploit the intermediate embeddings to train surrogate models to replace the bottom models on the clients (i.e., model stealing). The surrogate models can also be used to reconstruct private training data of the clients (i.e., data stealing). To defend against these stealing attacks, we propose Model Rake (i.e., Rake), which runs two bottom models on each client and differentiates their output spaces to make the two models distinct. Rake hinders the stealing attacks because it is difficult for a surrogate model to approximate two distinct bottom models. We prove that, under some assumptions, the surrogate model converges to the average of the two bottom models and thus will be inaccurate. Extensive experiments show that Rake is much more effective than existing methods in defending against both model and data stealing attacks, and the accuracy of normal model training is not affected.

As privacy concerns in recommender systems become increasingly prominent, federated recommender systems (FedRecs) have emerged as a promising distributed training paradigm.FedRecs enable the collaborative training of a shared global recommendation model without requiring the exchange of raw client interaction data.However, models trained using standard FedRec methods remain vulnerable to personal information leakage, particularly through attribute inference attacks, which can expose sensitive user attributes such as gender and race.In this paper, we address these user-sensitive attributes as targets for federated unlearning.To protect users' sensitive information, attribute unlearning aims to eliminate sensitive attributes from user embeddings, thereby preventing inference attacks while preserving recommendation performance.We introduce a novel post-training federated unlearning framework, Aegis, which performs unlearning based on private attribute requests after the model has been trained, minimizing the degradation in recommendation accuracy.Aegis employs an information-theoretic multi-component loss function to balance privacy protection and recommendation performance.Additionally, Aegis adapts to scenarios where training interaction data may be unavailable, reflecting real-world centralized protection scenarios.Comprehensive evaluations of various benchmark datasets demonstrate that our proposed method effectively safeguards user privacy while maintaining high-quality recommendations. CCS Concepts• Information systems → Social recommendation

With the rapid development of the Internet of Things (IoT), IoT devices find applications in various domains. The data generated by these devices is utilized for analysis and services, especially in the field of Artificial Intelligence (AI) applied to IoT, known as Artificial Intelligence of Things (AIoT). The enhancement of edge device computing power in the IoT has led to the emergence of research areas like edge-cloud synergy AI theories and application services. In the context of lifelong learning and real-time processes in AIoT edge-cloud synergy services, addressing unseen tasks becomes crucial. Unseen tasks arise when inference requests from edge devices involve models not present in the cloud’s model repository. Addressing these challenges involves generating data to either augment small sample problems or alter the data distribution for heterogeneous sample issues. As the application of large language models (LLMs) for data generation gains traction, challenges emerge in the context of AIoT edge-cloud synergy services. Firstly, fine-tuning LLMs with heterogeneous data exacerbates model bias issues. Secondly, the substantial data requirements for training LLMs pose a contradiction. Lastly, the involvement of manual annotation in LLM-based data generation introduces complexity and cost. This paper proposes a framework Seafarer to these challenges using Generative Adversarial Networks and Self-taught Learning. Seafarer avoids model bias, reduces data requirements, and eliminates the need for manual annotation. The design demonstrates effectiveness theoretically and is validated on the Cityscapes dataset, achieving an 80% reduction in training loss and improved validation loss stability.

Edge computing in federated learning based on centralized architecture often faces communication constraints in large clusters. Although there have been some efforts like computation-communication overlapping and fine-granularity flow scheduling towards how to reduce the communication cost, this is still a matter of ongoing research. Motivated by the underutilization of bandwidth among workers (edge devices) and the replication of deep neural network (DNN) model distributions in data-parallel federated learning, we propose OWL, a novel worker-assisted server bandwidth optimization method. OWL partitions numerous computation branches into groups based on the model's network topology , allowing for overlapping model distribution and computation among workers, thereby leveraging idle communication resources on the workers to compensate for server bandwidth. To address the issue of model distribution congestion on the server, we formulate group partition as an optimization problem , which proves to be NP-hard. We tackle this problem through a divide-and-conquer approach employing an approximation grouping algorithm and a deploying algorithm. Finally, we evaluate the performance of OWL through simulations and a comprehensive real-world case study involving model training on OWL and deployment on edge systems. Experimental results demonstrate that OWL reduces overall training time by up to 20%-69% and improves scalability by over 9.5% compared to state-of-the-art overlapping approaches.

In this paper, we propose pFedMo, a personalized federated learning algorithm with contrastive momentum. In pFedMo, we design a score function to personalize worker models by distilling knowledge from the aggregator's representation model so as to address the non-i.i.d. issue. To accelerate the convergence, we leverage the momentum acceleration on both the worker side and the aggregator side. However, the typical momentum without personalization does not suit well for the worker models with personalization, influencing convergence performance. To address this, we develop a personalized/contrastive momentum method for efficient momentum acceleration. We provide mathematical proof for the convergence of pFedMo on non-i.i.d. data. Extensive experiments based on real-world datasets and IoT system are conducted, verifying that pFedMo outperforms existing mainstream benchmarks, and achieves up to 35.90% accuracy increase and 3.64x training time speedup under a wide range of settings.

Finding the triangles of large scale graphs is a fundamental graph mining task in many applications, such as motif detection, microscopic evolution, and link prediction. The recent works on triangle counting can be classified into merge-based or binary search-based paradigms. The merge-based triangle counting paradigm locates the triangles using the set intersection operation, which suffers from the random memory access problem. The binary search-based triangle counting paradigm sets the neighbors of the source vertex of an edge as the lookup array and searches the neighbors of the destination vertex. There are lots of expensive lock operations needed in the binary search-based paradigm, which leads to low thread efficiency. In this paper, we aim to improve the triangle counting efficiency on GPU by designing a lock-free policy named Skiff to implement a hash-based triangle counting algorithm. In Skiff, we first design a hash trie data layout to meet the coalesced memory access model and then propose a lock-free policy to reduce the conflicts of the hash trie. In addition, we use a level array to manage the index of the hash trie to make sure the nodes of the hash trie can be quickly located. Furthermore, we implement a CTA thread organization model to reduce the load imbalance of the real-world graphs. We conducted extensive experiments on NVIDIA GPUs to show the performance of Skiff. The results show that Skiff can achieve a good system performance improvement than the state-of-the-art (SOTA) works.

Real-time video analytics demand intensive computing resources, often exceeding device capabilities. Heterogeneous computing resources like CPU and GPU, usually work collaboratively to ensure real-time performance. GPU manage data-intensive computing, while CPU handle instruction-intensive tasks. However, analytics accuracy can degrade because of the imbalance between CPU-GPU workloads and resources, and static resources in most systems limit adaptability to dynamic workloads. In addition to CPU-GPU resource control, accuracy is highly dependent on video analytics configuration including resolution, frame rate, and model selection. In this paper, we propose Gemini+, a hardware-accelerated video analytics pipeline empowered by dual-image FPGAs. It enables flexible CPU-GPU resource adaptation by providing near-instantaneous switching between two pre-configured images. We investigate CPU-GPU resource control and video analytics configuration adaptation in a dual-image FPGA-based video analytics pipeline. We study and formulate two problems of practical importance, a single-camera and multi-camera dual computing resource control problem, i.e., SC-DCRC and MC-DCRC. We analyze the problem complexity and develop optimal and sub-optimal algorithms. We evaluate our algorithms through simulation and build a prototype for verification. The results show that Gemini+ can improve analytic accuracy by 25% compared to fixed CPU-GPU resource systems and 88% compared to GPU-dominant systems.

Edge intelligence offers a promising solution for Simultaneous Localization and Mapping (SLAM) in large-scale scenarios, where multiple robots collaboratively perceive the environment and upload their local maps to an edge server. However, maintaining mapping accuracy under constrained and dynamic communication resources remains a significant challenge for the practical deployment of robot swarms. Concurrent data uploads from multiple agents can exacerbate network congestion, leading to the loss of critical information, delayed updates, and, ultimately, the inconsistency of the generated maps. This paper presents Hermes, an edge-assisted collaborative mapping system designed for communication-constrained environments. Hermes streamlines data transfer through bandwidth-aware map distillation, ensuring only the most crucial messages are transmitted to the edge server. We quantify the importance of keyframes and landmarks based on their information entropy gain in pose estimation. By selectively sharing essential submaps, Hermes adaptively balances communication bandwidth and information richness during the mapping process. We implemented Hermes on heterogeneous platforms and conducted experiments using public datasets and self-collected campus data. Hermes exceeds SwarmMap by 50% in bandwidth utilization with similar accuracy and surpasses COVINS-G by 65% in trajectory error under highly constrained network resources.

Local Differential Privacy (LDP) is a mechanism used to protect training privacy in Federated Learning (FL) systems, typically by introducing noise to data and local models. However, in real-world distributed edge systems, the non-independent and identically distributed nature of data means that clients in FL systems experience varying sensitivities to LDP-introduced noise. This disparity leads to fairness issues, potentially discouraging marginal clients from contributing further. In this paper, we explore how to enhance client-level performance fairness under LDP conditions. We model an FL system with LDP and formulate the problem PriFair using regularization, which assigns varied noise amplitudes to clients based on federated analytics. Additionally, we develop PriFairFed, a Tikhonov regularization-based algorithm that eliminates variable dependencies and optimizes variables alternately, while also offering a theoretical privacy guarantee. We further experimented with the algorithm on a real-world system with 20 Raspberry Pi clients, showing up to a 73.2% improvement in client-level fairness compared to existing state-of-the-art approaches, while maintaining a comparable level of privacy.

Federated Learning (FL) enables edge devices to collaboratively train a global model using local data. However, the increasing prevalence of watermarks in datasets presents a new challenge to efficient FL. While watermarks assert data ownership and copyright, they introduce complexities that can lead to shortcut learning problems and mislead utility measurements for client selection. These issues are further exacerbated by batch size variations in efficient FL frameworks, ultimately undermining their time-to-accuracy performance. We introduceLotusFL, an FL system designed to address the challenges posed by watermarked datasets in efficient FL. Specifically, it tackles the increased time-to-accuracy due to erroneous client selection and the accuracy degradation observed with larger batch sizes.LotusFLfirst estimates the characteristics of watermarks through statistical estimation and then adjusts the batch size using this estimated watermark information to balance the negative impact of the watermark against device idle waiting time. Additionally, its client selection mechanism, based on historical information, avoids the misleading utility signals from watermarks. This mechanism, working in conjunction with batch size adjustment, aims to accurately predict device runtime and identify potentially valuable devices. We evaluatedLotusFLthrough a real-world deployment on 40 edge devices. Compared to state-of-the-art efficient FL frameworks,LotusFLachieves superior performance, enhancing accuracy by up to 8.2% and reducing training time by 1.97×.

Federated unlearning (FU) has become an important area of research due to an increasing need for federated learning (FL) applications to comply with emerging data privacy regulations such as GDPR. It facilitates the removal of certain clients' data from an already trained FL model while preserving the performance on the remaining client without the need to retrain from scratch. Existing FU methods typically require clients to have access to their training data or historical model updates, which may be impractical in real-world scenarios due to privacy constraints and changes in data availability. Moreover, FU methods may cause catastrophic unlearning, where removing a client's data from heterogeneous, non-IID settings can negatively impact the model's performance on data from retained clients. To address the aforementioned issues and leverage the capabilities of personalized federated learning (pFL) in handling non-IID data distributions, this paper introduce Mimir, a novel data-free federated unlearning framework designed for pFL settings. Mimir integrates both learning and unlearning phases by utilizing personalized prompts for each client. We design a distillation structure based on Generative Adversarial Networks (GANs) for client-level unlearning that does not require access to original data or historical updates. By leveraging client-specific prompts generated during the pFL phase, Mimir adapts to heterogeneous data distributions and mitigates catastrophic unlearning on the retained data. We demonstrate the effectiveness of Mimir through extensive experiments on benchmark datasets, showing its ability to forget target client data while preserving model accuracy on the remaining clients.

Federated learning (FL) addresses privacy concerns by training models without sharing raw data, overcoming the limitations of traditional machine learning paradigms. However, the rise of smart applications has accentuated the heterogeneity in data and devices, which presents significant challenges for FL. In particular, data skewness among participants can compromise model accuracy, while diverse device capabilities lead to aggregation bottlenecks, causing severe model congestion. In this article, we introduce Spread+, a hierarchical system that enhances FL by organizing clients into clusters and delegating model aggregation to edge devices, thus mitigating these challenges. Spread+ leverages hedonic coalition formation game to optimize customer organization and adaptive algorithms to regulate aggregation intervals within and across clusters. Moreover, it refines the aggregation algorithm to boost model accuracy. Our experiments demonstrate that Spread+ significantly alleviates the central aggregation bottleneck and surpasses mainstream benchmarks, achieving performance improvements of 49.58% over FAVG and 22.78% over Ring-allreduce.

In the rapidly evolving landscape of cloud computing, serverless architectures offer a paradigm shift towards fine-grained function deployment and meticulous resource auto-scaling. Despite its growing popularity, existing systems often struggle to ensure the stability of function execution due to frequent cold starts and high concurrency demands. Our observations reveal a critical issue where a few hotspot functions excessively create new containers, resulting in substantial response latency fluctuations. To address this challenge, we propose Eunomia, a SLO-aware (Service Level Objective-aware) serverless framework. Eunomia introduces an optimized Poisson model with dynamic, sliding windows to accurately capture the arrival patterns of hotspot functions. Based on the optimized Poisson model, it proposes a queuing-based delay execution approach to mitigate initialization overhead by promoting instance reuse. Additionally, Eunomia designs flexible instance orchestration, providing dedicated concurrency pools for hotspot functions and dynamically adjusting the number of active instances. Experimental results demonstrate that Eunomia ensures 97% tail latency under a 100 ms response latency SLO, and outperforms the second-best baseline by 46% when memory is limited.

We present FedSM, a Federated Spectrum Management architecture to increase channel utilization (CU) and reduce latency, while protecting users’ data privacy. We employ hedonic coalition formation game for spectrum allocation. Within each coalition, we design a bandit learning algorithm to share spectra and adjust resource usage. Preliminary simulation results show FedSM increases CU to 93.51% and reduces latency to 248.68 ms compared to three privacy-preserving dynamic spectrum management architectures.

Personalized federated learning (PFL) has been proposed to overcome the challenge of statistical diversity in clients’ local data distribution under federated scenarios. Some existing PFL methods encourage collaboration between clients to improve overall accuracy. However, these methods often fail to consider the individual preferences of clients; instead, they adopt a top-down decision-making approach to divide clients into disjoint groups. In contrast, we consider the rationality of self-interested clients and allow them to make their own decisions regarding which federation to join. In our proposed PFL framework pFedMGC, we formulate the autonomous decision-making process of clients where they can participate in multiple federations. We then develop a heuristic algorithm to find feasible collaboration in this process. Clients in the same federations collaborate to train a federation-wide global model. To customize these federation models into clients’ personalized models, we employ an adaptive approach to calculate the aggregation weights of federation models based on differences in loss. The experiment results show that pFedMGC can improve overall personalized accuracy up to 28.89%, compared with the state-of-the-art PFL methods.

For graph self-supervised learning (GSSL), masked autoencoder (MAE) follows the generative paradigm and learns to reconstruct masked graph edges or node features while contrastive learning (CL) maximizes the similarity between augmented views of the same graph. Existing works utilize MAE and CL separately but we observe that the MAE and CL paradigms are complementary and propose the graph contrastive masked autoencoder (GCMAE) framework to unify them. Specifically, by focusing on local edges or node features, MAE cannot capture global information of the graph and is sensitive to particular edges and features. On the contrary, CL excels in extracting global information because it considers the relation between graphs. As such, we equip GCMAE with an MAE branch and a CL branch, and the two branches share a common encoder, which allows the MAE branch to exploit the global information extracted by the CL branch. To force GCMAE to capture global graph structures, we train it to reconstruct the entire adjacency matrix instead of only the masked edges as in existing works. Moreover, a discrimination loss is proposed for feature reconstruction, which improves the disparity between node embeddings rather than reducing the reconstruction error to tackle the feature smoothing problem of MAE. We evaluate GCMAE on four popular graph tasks (i.e., node classification, node clustering, link prediction, and graph classification) and compare it with 14 state-of-the-art baselines. The results show that GCMAE consistently provides good accuracy across these tasks, and the maximum accuracy improvement is up to 3.2% compared with the best-performing baseline.

Federated learning is a promising privacy-preserving learning paradigm in which multiple clients can collaboratively learn a model with their image data kept local. For protecting data ownership, personalized watermarks are usually added to the image data by each client. However, the introduced watermarks can lead to a shortcut learning problem, where the learned model performs predictions over-rely on the simple watermark-related features and represents a low accuracy on real-world data. Existing works assume the central server can directly access the predefined shortcut features during the training process. However, these may fail in the federated learning setting as the shortcut features of the heterogeneous watermarked data are difficult to obtain. In this paper, we propose a federated Morozov regularization technique, where the regularization parameter can be adaptively determined based on the watermark knowledge of all the clients in a privacy-preserving way, to eliminate the shortcut learning problem caused by the watermarked data. Specifically, federated Morozov regularization firstly performs lightweight local watermark mask estimation in each client to obtain the locations and intensities knowledge of local watermarks. Then, it aggregates the estimated local watermark masks to generate the global watermark knowledge with a weighted averaging. Finally, federated Morozov regularization determines the regularization parameter for each client by combining the local and global watermark knowledge. With the regularization parameter determined, the model is trained as normal federated learning. We implement and evaluate federated Morozov regularization based on a real-world deployment of federated learning on 40 Jetson devices with real-world datasets. The results show that federated Morozov regularization improves model accuracy by 11.22% compared to existing baselines.

With the rapid evolution of the Internet, Internet of Things (IoT), and geographic information systems (GIS), spatio-temporal Big Data (STBD) is experiencing exponential growth, marking the onset of the STBD era. Recent studies have concentrated on developing algorithms and techniques for the collection, management, storage, processing, analysis, and visualization of STBD. Researchers have made significant advancements by enhancing STBD handling techniques, creating novel systems, and integrating spatio-temporal support into existing systems. However, these studies often neglect resource management and system optimization, crucial factors for enhancing the efficiency of STBD processing and applications. Additionally, the transition of STBD to the innovative Cloud-Edge-End unified computing system needs to be noticed. In this survey, we comprehensively explore the entire ecosystem of STBD analytics systems. We delineate the STBD analytics ecosystem and categorize the technologies used to process GIS data into five modules: STBD, computation resources, processing platform, resource management, and applications. Specifically, we subdivide STBD and its applications into geoscience-oriented and human-social activity-oriented. Within the processing platform module, we further categorize it into the data management layer (DBMS-GIS), data processing layer (BigData-GIS), data analysis layer (AI-GIS), and cloud native layer (Cloud-GIS). The resource management module and each layer in the processing platform are classified into three categories: task-oriented, resource-oriented, and cloud-based. Finally, we propose research agendas for potential future developments.

In serverless computing, cold start results in long response latency. Existing approaches strive to alleviate the issue by reducing the number of cold starts. However, our measurement based on real-world production traces shows that the minimum number of cold starts does not equate to the minimum response latency, and solely focusing on optimizing the number of cold starts will lead to sub-optimal performance. The root cause is that functions have different priorities in terms of latency benefits by transferring a cold start to a warm start. In this paper, we proposeIncendio, a serverless computing framework exploiting priority-based scheduling to minimize the overall response latency from the perspective of cloud providers. We reveal the priority of a function is correlated to multiple factors and design a priority model based on Spearman’s rank correlation coefficient. We integrate a hybrid Prophet-LightGBM prediction model to dynamically manage runtime pools, which enables the system to prewarm containers in advance and terminate containers at the appropriate time. Furthermore, to satisfy the low-cost and high-accuracy requirements in serverless computing, we propose a Clustered Reinforcement Learning-based function scheduling strategy. The evaluations show that Incendio speeds up the native system by 1.4×, and achieves 23% and 14.8% latency reductions compared to two state-of-the-art approaches.

In this paper, we reference the previous conference version and complete the grant number mentioned in the acknowledgments of the conference version.

Consistent thickness in aluminum strips stands as a pivotal indicator of aluminum sheet product quality. Conventional automatic gauge control systems are complex, multivariable, and strongly coupled. However, the rolling process faces uncertainties, preventing the establishment of precise mathematical models. To tackle this, we propose an aluminum strip cold rolling thickness control method grounded in offline reinforcement learning. To facilitate the learning of better control policies, we construct a dataset of aluminum strip cold rolling process control data derived from real-world historical records and expertise for offline policy training, which is named dataset for aluminum strip cold rolling and comprises 8 373 540 Markov decision process tuples. We employ a clustered approach to handle time-varying production conditions. A constrained filtering scheme is introduced to eliminate problematic data after a data-driven ensemble rolling model is established. Evaluation and case study demonstrate that our method effectively reduces aluminum strip thickness deviations without requiring prior knowledge, thus improving control performance.

Generative Adversarial Networks (GANs) have become predominant in mobile computing for their ability to generate data. The concern for data privacy has made it arduous to collect large-scale datasets for GAN training on centralized servers. Federated Learning (FL) has emerged as a promising solution to address data privacy concerns. In this paper, we propose Oasis, a multiplayer-oriented federated GAN training system. We present a motivation, highlighting the Nash Equilibrium (NE) shift in vanilla federated GANs, exacerbated by data heterogeneity, leading to poor training performance with issues of vanishing gradient and mode collapse. To address mode collapse, Oasis extracts privacy-preserving data representations and generates a similarity table for clustering clients. Each group independently trains a GAN model and conducts distribution and fusion. By introducing a coordinator, Oasis generalizes intra-group games intoSeparable Zero-sum Multiplayer Gamesto tackle vanishing gradient. Thus, Oasis considers the overall federated GAN training asGroup-wise Separable Zero-sum Multiplayer Games. Practically, we evaluate our theoretical results both on a hardware prototype and in a simulated environment. Evaluation results demonstrate the effectiveness of Oasis, with an average improvement of 23.13% and 26.33% in terms of FID and NDB/K respectively, compared to threestate-of-the-artFL approaches over three datasets.

Recently, the rapid growth of video content and the increasing demand for high Quality of Experience (QoE) have significantly strained the backbone network. Edge caching is a promising approach to alleviate the strain by caching content closer to users. However, it confronts challenges stemming from the low capability of individual edge nodes and the high density of their distribution, resulting in low hit ratios and unbalanced workloads. In this paper, we conduct in-depth analyses of these challenges and formulate a typical cooperative edge caching problem. Based on the insights, we introduce MagNet, a cooperative edge caching system featuring two key mechanisms: Automatic Content Congregating (ACC) and Mutual Assistance Group (MAG). ACC improves hit ratios by intelligently guiding requests to their optimal edges, thereby facilitating content aggregation. Complementing this, Quick Cache is implemented to accelerate this congregation process by prefetching content and optimizing cache space, effectively boosting hit ratios. MAG, on the other hand, achieves workload balance by dynamically forming groups to augment edge capabilities and redistribute requests on overloaded edges. To elucidate the design principles of MagNet, we conduct detailed component-level comparisons and quantitative analyses. To validate the overall performance, we compare it with various caching solutions using real-world datasets, demonstrating significant performance improvements.

Scheduling and governing are two key technologies to trade off the Quality of Service (QoS) against the power consumption on mobile devices with heterogeneous cores. However, there are still defects in the use of them, among which two of the decoupling issues are critical and need to be resolved. First, both the scheduling and governing decouple from QoS, one of the most important metrics of user experience on mobile platforms. Second, scheduling and governing also decouple from each other in mobile systems and they might weaken each other when being effective at the same time. To address the above issues, we propose Orthrus, a comprehensive QoS-aware power management approach that involves a governing approach based on deep reinforcement learning to adjust the frequency of heterogeneous cores, a scheduling algorithm based on finite state machine that assigns cores to QoS-related threads, and expert fuzzy control-based coordination mechanism between the two to manage the impact between scheduling and governing. Our proposed approach aims to minimize power consumption while guaranteeing the QoS. We implement Orthrus on Google Pixel 3 as the system service of Android and evaluate it using several widespread mobile applications. The performance evaluation demonstrates that Orthrus reduces the average power consumption by up to 35.7% compared to three state-of-the-art techniques while ensuring the QoS on mobile platforms.

High Definition (HD) maps have become core supporting components for autonomous driving. To date, their updates heavily depend on the vehicle fleets of the map vendors, which cannot scale and timely reflect the highly dynamic environment. To ensure the HD map quality, it is advocated social vehicles should be used. Nevertheless, there are privacy concerns and a lack of incentives for social vehicles to contribute data. In this paper, we leverage federated analytics (FA), a newly developed collaborative data analytics paradigm, where raw data are kept local and only the insights generated from local analytics are sent to a server for aggregation. We present a new Federated Analytics based HD map Updating model (FAUMap) to protect the privacy of social vehicles. To motivate social vehicles to contribute data and improve the HD map quality, we formulate an overlapping coalition formation game, OCFUMap, and develop an algorithm to find feasible coalitions. Simulations show that our approach can improve the quality of the updated HD map by 1.56 times. To study an end-to-end operation of the FAUMap model and OCFUMap game, we present a case of HD map updates of the Powell street in San Francisco using the autonomous driving simulator CarLA.

Function-as-a-Service (FaaS) is a promising cloud computing model known for its scalability and elasticity. In various application domains, FaaS workflows have been widely adopted to manage user requests and complete computational tasks efficiently. Motivated by the fact that function containers collaboratively use the image layer's memory, co-placing functions would leverage memory sharing to reduce cluster memory footprint, this paper studies layer- wise memory sharing for serverless functions. We find that overwhelming memory sharing by placing containers in the same cluster machine may lead to performance deterioration and Service Level Objective (SLO) violations due to the increased CPU pressure. We investigate how to maximally reduce cluster memory footprint via layer- wise memory sharing for serverless workflows while guaranteeing their SLO. First, we study the container memory sharing problem under serverless workflows with a static Directed Acyclic Graph (DAG) structure. We prove it is NP-Hard and propose a 2-approximation algorithm, namely MDP. Then we consider workflows with dynamic DAG structure scenarios, where the memory sharing problem is also NP-Hard. We design a Greedy-based algorithm called GSP to address this issue. We implement a carefully designed prototype on the OpenWhisk platform, and our evaluation results demonstrate that both MDP and GSP achieve a balanced and satisfying state, effectively reducing up to 63% of cache memory usage while guaranteeing serverless workflow SLO.

Recently, Temporal Graph Neural Networks (TGNNs), as an extension of Graph Neural Networks, have demonstrated remarkable effectiveness in handling dynamic graph data. Distributed TGNN training requires efficiently tackling temporal dependency, which often leads to excessive cross-device communication that generates significant redundant data. However, existing systems are unable to remove the redundancy in data reuse and transfer, and suffer from severe communication overhead in a distributed setting. This work introduces Sven, a co-designed algorithm-system library aimed at accelerating TGNN training on a multi-GPU platform. Exploiting dependency patterns of TGNN models, we develop a redundancy-free graph organization to mitigate redundant data transfer. Additionally, we investigate communication imbalance issues among devices and formulate the graph partitioning problem as minimizing the maximum communication balance cost, which is proved to be an NP-hard problem. We propose an approximation algorithm called Re-FlexBiCut to tackle this problem. Furthermore, we incorporate prefetching, adaptive micro-batch pipelining, and asynchronous pipelining to present a hierarchical pipelining mechanism that mitigates the communication overhead. Sven represents the first comprehensive optimization solution for scaling memory-based TGNN training. Through extensive experiments conducted on a 64-GPU cluster, Sven demonstrates impressive speedup, ranging from 1.9x to 3.5x, compared to State-of-the-Art approaches. Additionally, Sven achieves up to 5.26x higher communication efficiency and reduces communication imbalance by up to 59.2%.

In recent years, the Mixture-of-Experts (MoE) technique has gained widespread popularity as a means to scale pretrained models to exceptionally large sizes. Dynamic activation of experts allows for conditional computation, increasing the number of parameters of neural networks, which is critical for absorbing the vast amounts of knowledge available in many deep learning areas. However, despite the existing system and algorithm optimizations, there are significant challenges to be tackled when it comes to the inefficiencies of communication and memory consumption. In this paper, we present the design and implementation of MPMoE, a high-performance library that accelerates MoE training with adaptive and memory-efficient pipeline parallelism. Inspired by that the MoE training procedure can be divided into multiple independent sub-stages. We design a pipeline parallelism method for reducing communication latency by overlapping with computation operations. Further, we analyze the memory footprint breakdown of MoE training and identify that activations and temporary buffers are the primary contributors to the overall memory footprint. Toward memory efficiency, we propose memory reuse strategies to reduce memory requirements by eliminating memory redundancies. Finally, to optimize pipeline granularity and memory reuse strategies jointly, we propose a profile-based algorithm and a performance model to determine the configurations of MPMoE at runtime. We implement MPMoE upon PyTorch and evaluate it with common MoE models in two physical clusters, including 64 NVIDIA A100 GPU cards and 16 NVIDIA V100 GPU cards. Compared with the state-of-art approach, MPMoE achieves up to 2.3× speedup while reducing more than 30% memory footprint for training large models.

With the increasing popularity of 3D volumetric video applications, e.g., metaverse, AR/VR, etc., there is a growing need to protect users' privacy while sharing their experiences during streaming. In this paper, we show that the existing privacy-preserving approaches for dense point clouds suffer a massive computation cost and degrade the quality of the streaming experience. We design Pagoda, a new PrivAcy-preservinG VOlumetric ViDeo StreAming incorporating the MPEG V-PCC standard, which protects different domain privacy information of dense point cloud, and maintains high throughput. The core idea is to content-aware transform the privacy attribute information to the geometry domain and content-agnostic protect the geometry information by adding Poisson noise perturbations. These perturbations can be denoised through a Poisson diffusion probabilistic model we design to deploy on the cloud. Users only need to encrypt a small amount of high-sensitive information and achieve secure streaming. Our designs ensure the dense point clouds can be transmitted in high quality and the attackers cannot reconstruct the original one. We evaluate Pagoda using three volumetric video datasets. The results show that Pagoda outperforms existing privacy-preserving baselines for 75.6% protection capability improvement, 4.27 times streaming quality, and 26 times latency reduction.

Internet of Things (IoT) gateways integrate various sensors and compute initial decisions before transmitting data to the cloud for further processing. As the functions they need to support become increasingly complex, gateways must upgrade their hardware. Network functions (NF) and video analytics (VAs) are two typical examples of hardware requirements: NFs need specialized hardware accelerators, while VAs need parallel processing power. However, gateways are typically constrained by factors, such as power, size, and cost, leading to a need to multiplex functions and minimize hardware overprovisioning. This article proposes a novel accelerator, the transmission-analytic processing unit (TAPU), which uses multi-image FPGA to accelerate VAs and NFs for IoT gateways. We preconfigure one image for VAs and one image for NFs, then multiplex the FPGA resources in the time dimension. The TAPU system design requires both hardware and software revisions. In the hardware design, we discuss our considerations on hardware choice and present a new abstraction of hardware functions to overcome the challenge of application development on different multi-image FPGAs. For the software, we develop a fully functional TAPU system to adapt to dynamic network and VAs workloads. Our evaluation shows that TAPU utilization can reach 92%, considerably increasing VAs and network processing throughput over the current approach. We further evaluate TAPU through two case studies that support a campus traffic monitoring system and an office surveillance system, demonstrating excellent performance improvement and low overhead.

Video analytics systems conduct video preprocessing to filter out unnecessary frames and model inference using appropriately selected neural networks for high analytics speed. Video preprocessing is instruction-intensive computing (IIC) executed by CPU, and model inference is data-intensive computing (DIC) executed by GPU. In this paper, we show the analytics accuracy of existing systems can largely vary in fields, caused by thedynamicIIC and DIC workloads of differentcontentsin applications. Unfortunately, cameras havefixedCPU/GPU resources and cannot effectively adapt to workload dynamics. We develop Gemini, a new edge-side real-time video analytics system enhanced by a dual-image FPGA. We take the advantage of negligible image switching time of dual-image FPGAs, pre-configure one CPU image and one GPU image and elastically multiplex the dual CPU-GPU resources intimedimension. Gemini requires both hardware and software revisions. In hardware, we overcome challenges of hardware-dependent application development, low communication efficiency between the microprocessor and FPGA, and high programming complexity by hardware abstraction, asynchronous data transfer mechanism and stub-skeleton middleware. In software, we overcome the challenge of adapting to the dynamic workloads by a bandit learning approach. We implement Gemini and show that Gemini can improve the analytics accuracy to 90.35%.

Recent advances in deep neural networks have substantially improved the accuracy and speed of various intelligent applications. Nevertheless, one obstacle is that DNN inference imposes a heavy computation burden on end devices, but offloading inference tasks to the cloud causes a large volume of data transmission. Motivated by the fact that the data size of some intermediate DNN layers is significantly smaller than that of raw input data, we designed the DNN surgery, which allows partitioned DNN to be processed at both the edge and cloud while limiting the data transmission. The challenge is twofold: (1) Network dynamics substantially influence the performance of DNN partition, and (2) State-of-the-art DNNs are characterized by a directed acyclic graph rather than a chain, so that partition is incredibly complicated. To solve the issues, We design a Dynamic Adaptive DNN Surgery(DADS) scheme, which optimally partitions the DNN under different network conditions. We also study the partition problem under the cost-constrained system, where the resource of the cloud for inference is limited. Then, a real-world prototype based on the selif-driving car video dataset is implemented, showing that compared with current approaches, DNN surgery can improve latency up to 6.45 times and improve throughput up to 8.31 times. We further evaluate DNN surgery through two case studies where we use DNN surgery to support an indoor intrusion detection application and a campus traffic monitor application, and DNN surgery shows consistently high throughput and low latency.

Local differential privacy (LDP) is a prominent approach and widely adopted in federated learning (FL) to preserve the privacy of local training data. It also nicely provides a rigorous privacy guarantee with computational efficiency in theory. However, a strong privacy guarantee with local differential privacy can degrade the adversarial robustness of the learned global model. To date, very few studies focus on the interplay between LDP and the adversarial robustness of federated learning. In this paper, we observe that LDP adds random noise to the data to achieve privacy guarantee of local data, and thus introduces uncertainty to the training dataset of federated learning. This leads to decreased robustness. To solve this robustness problem caused by uncertainty, we propose to leverage the promising distributionally robust optimization (DRO) modeling approach. Specifically, we first formulate a distributionally robust and private federated learning problem (DRPri). While our formulation successfully captures the uncertainty generated by the LDP, we show that it is not easily tractable. We thus transform our DRPri problem to another equivalent problem, under the Wasserstein distance-based uncertainty set, which is named the DRPri-W problem. We then design a robust and private federated learning algorithm, RPFL, to solve the DRPri-W problem. We analyze RPFL and theoretically show it satisfies differential privacy with a robustness guarantee. We evaluate algorithm RPFL by training classifiers on real-world datasets under a set of well-known attacks. Our experimental results show our algorithm RPFL can significantly improve the robustness of the trained global model under differentially private data by up to 4.33 times.

Federated learning (FL) is a new distributed machine learning paradigm that enables machine learning on edge devices. One unique feature of FL is that edge devices belong to individuals; and since they are not “owned” by the FL coordinator, but can be “federated” instead, there can potentially be a huge number of edge devices. In the current distributed ML architecture, the parameter server (PS) architecture, model aggregation is centralized. When facing a large number of edge devices, the centralized model aggregation becomes the bottleneck and fundamentally restricts system scalability.

Edge-side real-time video analytics systems recognize spatial or temporal events (e.g., vehicle counting) in a video stream. To meet the delay requirement, existing systems in smart edge cameras conduct video preprocessing to filter out unnecessary frames and model inference using appropriately selected neural network (NN) models. Video preprocessing is instruction-intensive computing (IIC) and executed by the CPU of the edge camera, and model inference is data-intensive computing (DIC) and executed by the GPU of the edge camera. In this paper, we show that the analytics accuracy of existing systems can largely vary in fields. The root cause is that video analytics applications have different contents, which result in dynamic IIC and DIC workloads. Unfortunately, intelligent cameras in fields have fixed CPU and GPU resources and cannot effectively adapt to workload dynamics. We develop Gemini, a new real-time video analytics system enhanced by a dual-image FPGA. The newly developed dual-image FPGAs can be pre-configured with two FPGA images with a key advantage of negligible image switching time. We thus pre-configure one CPU image and one GPU image and elastically multiplex the dual CPU-GPU resources in the time dimension. The Gemini system design requires both hardware and software revisions. We overcame a challenge that the application development on different dual-image FPGAs is hardware-dependent. We develop a new abstraction of hardware functions to make the Gemini system hardware-agnostic. It is also a challenge to adapt to the dynamic workloads and optimize video analytics accuracy. We develop a bandit learning approach to capture content dynamics and conduct dual computing resource control. We implement Gemini and show that Gemini can improve the analytics accuracy to 90.35 %. We further evaluate Gemini by a case study where we use Gemini to support an intrusion detection application, and Gemini shows consistent high analytics accuracy.

Real-time edge-cloud video analytics systems have been widely used to support such applications as traffic counting, surveillance, autonomous driving, Metaverse, etc. In such a system, the edge and the cloud cooperatively conduct model inference of the video frames captured by the camera of the edge, using a trained DNN model of the video analytics application. The edge conducts initial analytics on the video frames to a split layer of the DNN model; and then sends intermediate results to the cloud for follow-up analytics. In this paper, we show that an attacker can perform reconstruction attacks to the intermediate results; and private information of the raw video frames, e.g., a plate number of a car, can be leaked. In this paper, we present Preva, a new Privacy preserving Real-time Edge-cloud Video Analytics system. The core idea of Preva is to conduct image transformation on the video frames, as preprocessing, prior to the video frames starting the edge-cloud video analytics process, so that during edge-cloud video analytics, the intermediate results will not leak private information under attack. We design a policy-based video-frame transformation scheme. Given the resource constraints of the edge, Preva ensures high accuracy in the final video analytics results and minimizes privacy leakage in any split layer. We present a formal privacy analysis and we show that Preva can guarantee privacy leakage under the reconstruction attacks of both outsider attackers and insider attackers. We evaluate Preva through three video analytics applications and we show that Preva outperforms existing systems for 64.4% in analytics accuracy and 59.2% in privacy leakage.

Nowadays, the surge of Internet contents and the need for high Quality of Experience (QoE) put the backbone network under unprecedented pressure. The emerging edge caching solutions help ease the pressure by caching contents closer to users. However, these solutions suffer from two challenges: 1) a low hit ratio due to edges’ high density and small coverages. 2) unbalanced edges’ workloads caused by dynamic requests and heterogeneous edge capacities. In this paper, we formulate a typical cooperative edge caching problem and propose the MagNet, a decentralized and cooperative edge caching system to address these two challenges. The proposed MagNet system consists of two innovative mechanisms: 1) the Automatic Content Congregating (ACC), which utilizes a neural embedding algorithm to capture underlying patterns of historical traces to cluster contents into some types. The ACC then can guide requests to their optimal edges according to their types so that contents congregate automatically in different edges by type. This process forms a virtuous cycle between edges and requests, driving a high hit ratio. 2) the Mutual Assistance Group (MAG), which lets idle edges share overloaded edges’ workloads by forming temporary groups promptly. To evaluate the performance of MagNet, we conduct experiments to compare it with classical, Machine Learning (ML)-based and cooperative caching solutions using the real-world trace. The results show that the MagNet can improve the hit ratio from 40% and 60% to 75% for non-cooperative and cooperative solutions, respectively, and significantly improve the balance of edges’ workloads.

The local model poisoning attack is an attack to manipulate the shared local models during the process of distributed learning. Existing defense methods are passive in the sense that they try to mitigate the negative impact of the poisoned local models instead of eliminating them. In this paper, we leverage the new federated analytics paradigm, to develop a proactive defense method. More specifically, federated analytics is to collectively carry out analytics tasks without disclosing local data of the edge devices. We propose a Federated Anomaly Analytics enhanced Distributed Learning (FAA-DL) framework, where the clients and the server collaboratively analyze the anomalies. FAA-DL firstly detects all the uploaded local models and splits out the potential malicious ones. Then, it verifies each potential malicious local model with functional encryption. Finally, it removes the verified anomalies and aggregates the remaining to produce the global model. We analyze the FAA-DL framework and show that it is accurate, robust, and efficient. We evaluate FAA-DL by training classifiers on MNIST and Fashion-MNIST under various local model poisoning attacks. Our experiment results show FAA-DL improves the accuracy of the learned global model under strong attacks up to 6.90 times and outperforms the state-of-the-art defense methods with a robustness guarantee.

On edge devices, data scarcity occurs as a common problem where transfer learning serves as a widely-suggested remedy. Nevertheless, transfer learning imposes heavy computation burden to the resource-constrained edge devices. Existing task allocation works usually assume all submitted tasks are equally important, leading to inefficient resource allocation at a task level when directly applied in Multi-task Transfer Learning (MTL). To address these issues, we first reveal that it is crucial to measure the impact of tasks on overall decision performance improvement and quantify task importance. We then show that task allocation with task importance for MTL (TATIM) is a variant of NP-complete Knapsack problem, where the complicated computation to solve this problem needs to be conducted repeatedly under varying contexts. To solve TATIM with high computational efficiency, we propose a Data-driven Cooperative Task Allocation (DCTA) approach. Finally, we evaluate the performance of DCTA by not only a trace-driven simulation, but also a new comprehensive real-world AIOps case study which bridges model and practice via a new architecture and main components design within AIOps system. Extensive experiments show that our DCTA reduces 3.24 times of processing time, and saves 48.4 percent energy consumption compared with the state-of-the-art when solving TATIM.

Presents corrections to author information for the above named paper.

Edge computing for machine learning has become a heated research topic. On edge devices, data scarcity occurs as a common problem where transfer learning serves as a widely-suggested remedy. Nevertheless, one obstacle is that transfer learning imposes heavy computation burden to the resource-constrained edge devices. Motivated by the fact that only a few tasks of Multi-task Transfer Learning (MTL) have a higher potential for overall decision performance improvement, we design a novel task allocation scheme, which assigns more important tasks to more powerful edge devices to maximize the overall decision performance. In this paper, we focus on task allocation under multi-task scenarios by introducing task importance and make the following contributions. First, we reveal that it is important to measure the impact of tasks on overall decision performance improvement and quantify task importance. We also observe the long-tail property of task importance, i.e., only a few tasks are important, which facilitates more efficient task allocation. Second, we show that task allocation with task importance for MTL (TATIM) is in fact a variant of the NP-complete Knapsack problem, where the complicated computation to solve this problem needs to be conducted repeatedly under varying contexts. To solve TATIM with high computational efficiency, we innovatively propose a Data-driven Cooperative Task Allocation (DCTA) approach. Third, we evaluate the performance of our DCTA approach by applying it to a real-world industrial operation (e.g., AIOps) scenario. Experiments show that our DCTA approach can reduce 3.24 times of processing time compared with the state-of-the-art when solving TATIM. We offer our DCTA approach as an effective and practical mechanism for reducing the required resource associated with performing MTL on edge devices.

Recent advances in deep neural networks (DNNs) have substantially improved the accuracy and speed of a variety of intelligent applications. Nevertheless, one obstacle is that DNN inference imposes heavy computation burden to end devices, but offloading inference tasks to the cloud causes transmission of a large volume of data. Motivated by the fact that the data size of some intermediate DNN layers is significantly smaller than that of raw input data, we design the DNN surgery, which allows partitioned DNN processed at both the edge and cloud while limiting the data transmission. The challenge is twofold: (1) Network dynamics substantially influence the performance of DNN partition, and (2) State-of-the-art DNNs are characterized by a directed acyclic graph (DAG) rather than a chain so that partition is greatly complicated. In order to solve the issues, we design a Dynamic Adaptive DNN Surgery (DADS) scheme, which optimally partitions the DNN under different network condition. Under the lightly loaded condition, DNN Surgery Light (DSL) is developed, which minimizes the overall delay to process one frame. The minimization problem is equivalent to a min-cut problem so that a globally optimal solution is derived. In the heavily loaded condition, DNN Surgery Heavy (DSH) is developed, with the objective to maximize throughput. However, the problem is NP-hard so that DSH resorts an approximation method to achieve an approximation ratio of 3. Real-world prototype based on self-driving car video dataset is implemented, showing that compared with executing entire the DNN on the edge and cloud, DADS can improve latency up to 6.45 and 8.08 times respectively, and improve throughput up to 8.31 and 14.01 times respectively.

Nowadays, manufacturers want to collect the data of their sold-products to the cloud, so that they can conduct analysis and improve the operation, maintenance and services of their products. Manufacturers are looking for a self-contained solution for data transmission since their products are typically deployed in a large number of different buildings, and it is neither feasible to negotiate with each building to use the building's network (e.g., WiFi) nor practical to establish its own network infrastructure. ISPs are aware of this market. Since the readily available 3G/4G is over costly for most IoT devices, ISPs are developing new choices. Nevertheless, it can be expected that the choices from ISPs will not be fine-grained enough to match hundreds or thousands of requirements on different costs and data volumes from the IoT applications. To address this problem, we for the first time propose IoT communication sharing (ICS). We first clarify the ICS scenarios. We then formulate the IoT communication sharing (ICS) problem, and develop a set of algorithms with provable performance. We further present our implementation of a fully functioning system. Our evaluations show that ICS and our algorithms can lead to a cost reduction of five times and eight times respectively for the two real-world cases.

Nowadays, manufacturers want to send the data of their products to the cloud so that they can conduct analysis and improve their operation, maintenance, and services. Manufacturers are looking for a self-contained solution. This is because their products are deployed in a large number of different buildings, and it is neither feasible for a vendor to negotiate with each building to use the building’s network (e.g., WiFi) nor practical to establish its own network infrastructure. The vendor can rent a dedicated channel from an ISP to act as a thing-to-cloud communication (TCC) link for each of its IoT devices. The readily available choices, e.g., 3G, is over costly for most IoT devices. ISPs are developing cheaper choices for TCC links, yet we expect that the number of choices for TCC links will be small as compared to hundreds or thousands of requirements on different costs and data rates from IoT applications. We address this issue by proposing a communication sharing architecture sTube+, sharing tube . The objective of sTube+ is to organize a greater number of IoT devices, with heterogeneous data communication and cost requirements, to efficiently share fewer choices of TCC links and transmit their data to the cloud. We take a design of centralized price optimization and distributed network control. More specifically, we architect a layered architecture for data delivery, develop algorithms to optimize the overall monetary cost, and prototype a fully functioning system of sTube+. We evaluate sTube+ by both experiments and simulations. In addition, we develop a case study on smart maintenance of chillers and pumps, using sTube+ as the underlying network architecture.

Data replication is widely used in geo-distributed storage systems, and strong consistency is preferred for correctness and programming simplicity at the application layer. To address the inefficiency and insufficiency of the causal consistency model, a strong consistency model named distributed context consistency is defined. It explicitly defines the necessary dependencies among distributed clients to effectively reduce falsepositive dependencies among operations. A consensus algorithm named CC-Paxos is proposed to implement this distributed context consistency model. It exploits timestamps for operation sequencing in distributed contexts and adopts fine-granularity dependency checking to effectively reduce the number of potential conflicts. Experimental results show that, compared with implementations using causal+ consistency model in the upper layer and Egalitarian Paxos in system layer, CC-Paxos can significantly decreases latency and increases throughput with no sacrifice on scalability.

Cloud-based file systems are widely accepted and adopted for personal and business purposes in recent years. Statistics shows that ∼25% of file operations from a typical user are random writes. Inherited from traditional disk-based file systems, most distributed file systems are also based on objects or chunks of fixed sizes, which work well for sequential writes but poorly for random writes. This paper investigates the design paradigm of variable-sized objects for a distributed file system, where a new file write interface is proposed to provide rich write semantics. A novel distributed file system named VarFS, is presented to incorporate variable object indexing, support the random write interface and remain POSIX compatible. VarFS reduces the amount of unnecessary data being read and the number of objects modified in face of updates and consequently alleviates the total amount of data transferred. VarFS is implemented based on Ceph and the performance measurements show that it can achieve 1–2 orders of magnitude less latency than Ceph on random writes. At the same time, the overhead for initial writes and re-writes is acceptable.