Virtualization environments (e.g., containers and hypervisors) achieve isolation of multiple runtime entities but result in two mutually isolated guest and host layers. Such cross-ayer isolation could cause high latency and low throughput of the system. Previous aware scheduling and double scheduling fail to achieve bidirectional coordination between the guest and host layers. To address this challenge, we develop DCS3, a Dual-layer Co-aware Scheduler that combines stealing balance and synchronized priority. Stealing balancing migrates tasks between virtual CPU (vCPU) queues for load balance based on the workloads of physical CPUs (pCPUs). Synchronized priority dynamically adjusts the thread priorities running on the pCPUs according to the current vCPU workloads. The vCPUs and pC-PUs belong to the guest and host layers, respectively. Compared with aware scheduling, double scheduling, and DCS2 (i.e., DCS3 without synchronized priority), DCS3 has the following obvious advantages: 1) Requests Per Second (RPS) increases by up to 52%, 55%, and 2%, respectively; 2) request latency decreases by up to 72%, 71%, and 20%, respectively.

With the rapid development of deep learning techniques in mobile and embedded devices, light-weight inference engines (e.g., Paddle Lite and TensorFlow Lite) are emerged. In some real-time application scenarios, these light-weight inference engines require time acceleration and low memory consumption. Paddle Lite is a well-known open-source inference engine that is fully functional. However, Paddle Lite only supports regular OS (e.g., Linux, Windows, and iOS), making it difficult to achieve time acceleration and low memory consumption for real-time application scenarios during inference. In this brief, we propose the Paddle Lite on Zephyr solution for inference acceleration in RTOS. We first propose a modular compilation method to incorporate the most basic functions of Paddle Lite. To address the system differences between RTOS and Linux, we resolve the system-level and compilation-level issues from modular compilation. We then load the Paddle Lite model into memory as a device when the system starts up. We further design an inference method that skips third-party libraries during inference and thus obtains the same inference results as Linux. We deploy the Paddle Lite on Zephyr and conduct experiments with seven classic Convolutional Neural Network (CNN) models on a single-core CPU. The experiment results show that the average inference time on Zephyr RTOS is reduced by 7%, and the average memory consumption is reduced by 78% compared to Linux. This work has merged an upstream branch of the Paddle Lite.

Multimodal recommendation is a crucial technology on social media platforms. It is widely applied in scenarios such as product recommendation and advertising delivery. However, existing multimodal recommendation approaches often overlook invariant semantic features that persist across modalities, leading to decreased robustness and generalization. To address this limitation, we proposeGILMRec, a novelgraphinvariantlearning-basedmultimodal social mediarecommendation framework. The GILMRec introduces an invariant feature learning strategy to extract invariant features separately from visual and textual modalities and employs an attention-based fusion mechanism to integrate them into a unified embedding. Specifically, we construct modality-specific similarity graphs and apply top-$t$neighbor aggregation, enhancing the consistency of invariant features while effectively suppressing modality-specific noise. Extensive experiments on three Amazon benchmark datasets and a large-scale dataset [baby, sports, clothing, and compact discs (CDs)] demonstrate that GILMRec consistently outperforms twelve state-of-the-art baselines. The results confirm the efficiency of invariant features in capturing robust multimodal representations and improving recommendation performance, particularly in sparse data scenarios.

Self-supervised heterogeneous graph representation learning (SSHGRL) is a key technique for embedding heterogeneous graphs, enabling effective analysis and modeling of social networks and other graph-structured data, which are central to knowledge discovery and the study of social systems. However, existing SSHGRL methods are hardly applied to large-scale heterogeneous graph environments due to the normally used metapath decomposing mechanism being graph-size-sensitive. Moreover, the existing self-supervised signals are normally created from Shared Mutual Information (SMI) of different graph views that ignore the Non-SMI (NMI) contained in the same view. This results in the model tending to learn insufficient graph representation. To this end, this article proposes a designated masking propagation (DMP) mechanism to process heterogeneous graphs without using metapath. Moreover, based on the DMP graph view, a novel sufficient representation is proposed to learn the effective graph representation by combining both NMI and SMI. Extensive experiments on eight large- and medium-scale heterogeneous graph datasets demonstrate the superiority of our method, setting new state-of-the-art performance in various big data contexts.

Blockchain-enabled mobile edge computing (MEC) must jointly optimizetask offloadingandconsensus finalityunder highly heterogeneous AIoT devices, where latency/energy constraints and fairness-sensitive incentives coexist with time-varying validator reliability. We proposeFE-CTDE, a unified framework that couples (1) a Stackelberg pricing-and-allocation layer that reaches a unique equilibrium and reduces utility disparity, (2) a reliability-aware dynamic BFT committee and block-packing mechanism that stabilizes confirmation delay under intermittent connectivity, and (3) a centralized-training/decentralized-execution multi-agent policy that outputs a continuous offloading ratio while requiring only local observations at run time. Extensive simulations across diverse heterogeneity, workload burstiness, and link intermittency show that FE-CTDE consistently improves social welfare and fairness while reducing end-to-end latency/energy and sustaining highereffectiveconsensus throughput, outperforming strong baselines by up to22.23%. We further report protocol/learning overheads and provide reproducible implementation details.

Distributed Computing Continuum Systems (DCCS) unify cloud, fog, edge, and Internet of Things (IoT) into a single execution fabric. At scale, heterogeneity and bursty arrivals make it hard to meet per task deadlines. Prior approaches based on learning, optimization, market mechanisms, or class level slicing depend on global state or iterative coordination. Decisions lag arrivals, isolation is scoped to coarse classes rather than individual tasks, and the deadline violation ratio (DVR) rises. We present Task-level Adaptive Computing Slicing (TACS), a fine grained slicing paradigm that delivers task aligned resource governance through autonomous shard management. For each arriving task, TACS executes decentralized scheduling at the shard level to instantiate an ephemeral micro-slice governed by an autonomous shard formed exactly by the task's participants. Within the shard, participants apply closed form rules to make local resource allocation decisions. This design achieves strict task level isolation and precise, scalable matching between supply and demand without global coordination or model retraining. In simulations with 100 to 1000 heterogeneous nodes and a range of loads and heterogeneity levels, TACS maintains DVRs below 1% and provides steadier, higher throughput than advanced baselines. Under adverse conditions, representative baselines exceed 20% DVR and in several cases require retraining when device populations change.

The convergence of federated learning (FL) and blockchain in edge-end-cloud systems offers promising opportunities for privacy-preserving collaborative intelligence. However, existing blockchain-enhanced FL (BFL) approaches remain vulnerable to malicious participants and lack robust protection for model updates. To address these issues, we propose SEPP-FLBC, a Secure and Efficient Privacy Protection framework based on Federated Learning and Blockchain Committees. SEPP-FLBC introduces a novel blockchain committee consensus mechanism to validate model updates and defend against unreliable nodes. It further employs a refined multi-party communication paradigm to facilitate indirect and secure data interactions, reducing the risk of information leakage. Additionally, differential privacy noise is applied to model updates to enhance resistance to inference attacks. A formal convergence analysis is conducted to ensure model stability and minimize overhead. Extensive experiments on benchmark datasets demonstrate that SEPP-FLBC achieves superior accuracy while maintaining strong privacy guarantees and communication efficiency, outperforming state-of-the-art BFL methods in both security and performance.

In mission-critical IoT-MEC environments, jointly optimizing service placement and resource allocation is intractable due to the high-dimensional coupling of discrete topological decisions with continuous resource dimensioning. Furthermore, traditional methods oversimplify dependencies, overlooking multi-source “Wait-for-All” synchronization and the stochastic variance of bursty workloads. To bridge these gaps, we propose HyDK, a variance-aware framework synergizing Deep Reinforcement Learning (DRL) with convex optimization. The core innovation is our Action Space Pruning mechanism. We theoretically decompose the hybrid decision space by solving the continuous sub-problem to optimality via a KKT-based convex optimization routine. This acts as a deterministic optimality backstop, effectively pruning continuous dimensions and allowing the agent to focus exclusively on the complex discrete topological search. To address physical realities, we construct a finegrained Directed Acyclic Graph (DAG) model to capture data aggregation bottlenecks and integrate an M/G/1 queuing model incorporating the second moment of service time to mitigate longtail latency risks. Trace-driven simulations using the Edge-IIoTset demonstrate that HyDK improves system responsiveness by up to 25.6% with significantly tighter confidence intervals compared to existing baselines.

Anomalies often occur in real-world information networks/graphs, such as malevolent users, malicious comments, banned users, and fake news in social graphs. The latest graph anomaly detection methods use a novel mechanism called truncated affinity maximization (TAM) to detect anomaly nodes without using any label information and achieve impressive results. TAM maximizes the affinities among the normal nodes while truncating the affinities of the anomalous nodes to identify the anomalies. However, existing TAM-based methods truncate suspicious nodes according to a rigid threshold that ignores the specificity and high-order affinities of different nodes. This inevitably causes inefficient truncations from both normal and anomalous nodes, limiting the effectiveness of anomaly detection. To this end, this paper proposes a novel truncation model combining contextual and global affinity to truncate the anomalous nodes. The core idea of the work is to use contextual truncation to decrease the affinity of anomalous nodes, while global truncation increases the affinity of normal nodes. Extensive experiments on massive real-world datasets show that our method surpasses peer methods in most graph anomaly detection tasks. In highlights, compared with previous state-of-the-art methods, the proposed method has +15% ~ +20% improvements in two famous real-world datasets, Amazon and YelpChi. Notably, our method works well in large datasets, Amazin-all and YelpChi-all, and achieves the best results, while most previous models cannot complete the tasks.

Generalist Graph Anomaly Detection (GGAD) extends traditional Graph Anomaly Detection (GAD) from one-for-one to one-for-all scenarios, posing significant challenges due to Feature Space Shift (FSS) and Graph Structure Shift (GSS). This paper first formalizes these challenges and proposes quantitative metrics to measure their severity. To tackle FSS, we develop an anomaly-driven graph invariant learning module that learns domain-invariant node representations. To address GSS, a novel structure-insensitive affinity learning module is introduced, capturing cross-domain structural correspondences via affinity-based features. Our unified framework, IA-GGAD, integrates these modules, enabling anomaly prediction on unseen graphs without target-domain retraining or fine-tuning. Extensive experiments on benchmark datasets from varied domains demonstrate IA-GGAD’s superior performance, significantly outperforming state-of-the-art methods (e.g., achieving up to +12.28\% AUROC over ARC on ACM). Ablation studies further confirm the effectiveness of each proposed module. The code is available at \url{https://github.com/kg-cc/IA-GGAD/}.

Blockchain technology underpins the value internet, yet the isolation of blockchain systems creates "data and value islands," limiting interoperability. To address this challenge, we propose a scalable and secure cross-chain framework that eliminates reliance on relay chains and enhances performance through decentralized execution and state sharding. Each business blockchain operates as an autonomous node, collaboratively maintaining a virtual cross-chain transaction blockchain. Our framework achieves significant improvements: when state operations are executed on-chain, the framework(with a block size of 2) delivers 66% and 149% higher transactions/s compared to TCIP and BitXhub, respectively, with 20% lower latency. Furthermore, off-chain execution(with a block size of 2) boosts maximum transactions/s by 63% and reduces latency by 36%. The introduction of state sharding enhances parallel transaction execution(with a block size of 128 and a number of sharding nodes of 16), achieving an additional 108% performance improvement. This work demonstrates a novel approach to cross-chain interoperability, offering a secure, efficient, and scalable solution for complex blockchain ecosystems.

While directed acyclic graph (DAG) blockchain technologies improve scalability and throughput over traditional blockchains, they still face critical challenges, particularly in efficiently determining transaction order and ensuring verifiable transaction selection. These limitations significantly hinder their applicability in high-demand environments like IoT, where secure, high-throughput processing is essential. To address these issues, we propose a DAG-partitioned multichain architecture (DPMA), which generates distinct chains for each user node, enabling parallel transaction processing. In addition, we introduce a transaction selection algorithm with verifiable random function (TSAV), which implements a verifiable two-tier transaction selection process, ensuring secure and transparent transaction ordering. To further enhance security, we propose a dynamic transaction confidence analysis method that adjusts VRF parameters in response to network conditions. Experimental results demonstrate that our approach effectively identifies malicious behavior and improves transaction credibility. Compared to existing methods, our solution offers enhanced scalability and security, making it well-suited for large-scale decentralized applications.

Hypervisor is a VMM (Virtual Machine Monitor) that creates and runs multiple VMs (Virtual Machines) through abstracting resources from a physical machine. Hypercall is a special and crucial call used in virtualized systems as it serves as a main communication channel between VMs and the hypervisor. However, hypercall attacks occur when an attacker manipulates the communication channel, and it could cause abnormal VM status, potentially leading to the abnormal resource allocation of the host OS (Operating System) and crash of VMs. Therefore, the virtualized system should execute abnormal VM status detection to identify potential abnormal behaviors to protect VMs and the host OS; however, existing works are either for reconstructing the hypervisor or hardware isolation, not for the VM status detection for abnormal hypercall.This study develops a hypercall-oriented abnormal VM status detection system called HypercallDetector based on the following three innovations: 1) we implement a hypercall tracing based on eBPF to obtain the hypercall-related running status (including CPU usage, memory usage, network traffic, etc.) of each VM; 2) we implement a window division technology to divide the VM status into multiple status windows of the same size, and appropriate window size with balanced detection precision (95.0%) and latency (within 8.8 ms) obtained by proposing the window regulator; and 3) we implement a CS-H algorithm (Compressing Sensing for Hypercall) to distinguish whether the VM status is abnormal. HypercallDetector shows higher precision and lower latency than its opponent and consumes only 8.6% CPU of single core and 0.3% memory usage when starting 240 VMs.

Training large Deep Convolutional Neural Networks (DCNNs) with increasingly large datasets to improve model accuracy has become extremely time-consuming. Distributed training methods, such as data parallelism (DP) and pipeline model parallelism (PMP), offer potential solutions but face challenges like load imbalance and significant communication overhead. This paper introduces GroPipe, a novel architecture that synergistically integrates PMP and DP, markedly improving training speeds. GroPipe employs an automatic model partitioning algorithm based on a performance projection technique, ensuring load balance and facilitating quantitative performance evaluation in PMP. Additionally, it adopts a group-based delayed asynchronous communication strategy to efficiently reduce communication overhead in DP. Using the ResNet and VGG models with the ImageNet dataset, extensive experiments are performed on an 8-GPU server and demonstrate GroPipe’s effectiveness. GroPipe achieves substantial improvements in time to accuracy, showing an average improvement of 42.2% and 14.0% on the ResNet series, and 79.2% and 43.9% on the VGG series, without compromising Top-1 accuracy.

Mobile-Edge Cloud Computing (MECC) plays a crucial role in balancing low-latency services at the edge with the computational capabilities of cloud data centers (DCs). However, many existing studies focus on single-provider settings or limit their analysis to interactions between mobile devices (MDs) and edge servers (ESs), often overlooking the competition that occurs among ESs from different providers. This article introduces an innovative two-layer game framework that captures independent self-interested competition among MDs and ESs, providing a more accurate reflection of multi-vendor environments. Additionally, the framework explores the influence of cloud-edge collaboration on ES competition, offering new insights into these dynamics. The proposed model extends previous research by developing algorithms that optimize task offloading and resource allocation strategies for both MDs and ESs, ensuring the convergence to Nash equilibrium in both layers. Simulation results demonstrate the potential of the framework to improve resource efficiency and system responsiveness in multi-provider MECC environments.

The distance vector hop algorithm is commonly used for sensor node localization. However, its high localization accuracy error and stability issues make it unsuitable for many applications. To overcome these concerns, this article proposes a new function and binary distance vector hop (FBDV-Hop) algorithm with binary controllers and function correction methods while considering the application requirement. In FBDV-Hop, binary controllers are designed to analyze the optimization effect of the module fully and make it adaptable to diverse scenarios. The correction strategies were based on average hop distance measurement, estimated distance, equation composition method, and localization after supplementary correction, which were divided into four modules in accordance with the module error sources of different design correction functions. The simulation experiment was designed to analyze the principle of the role of each module in depth and achieve the optimal optimization effect. The experimental results show that the localization error optimization rate under the FBDV-Hop algorithm was more than 70%, and the optimization rate, stability, effectiveness, and adaptability of the algorithm were considerably greater than the baseline algorithms.

UAV-assisted MEC revolutionizes edge computing by deploying UAVs for real-time data processing in areas lacking infrastructure, supporting a wide range of applications from emergency responses to smart cities. Unlike edge servers, UAVs face substantial computational constraints, necessitating a comprehensive strategy that integrates UAV trajectory with task offloading, caching, and migration. Existing studies often overlook the synergy among these strategies, impacting their overall effectiveness. Furthermore, the focus on content pre-caching overlooks task caching’s critical role in addressing high computational demands with limited UAV resources. This research aims to jointly optimize UAV trajectories and task management strategies, including offloading, caching, and migration. Utilizing the Lyapunov optimization framework, we break down the complex optimization problem into manageable subproblems: UAV placement, user-UAV association, task offloading, scheduling, and bandwidth allocation, addressed iteratively using the Block Coordinate Descent method. Specifically, the scheduling subproblem is transformed into a non-convex quadratically constrained quadratic programming problem, managed effectively through semidefinite relaxation and a probabilistic mapping approach. Our simulations show that this integrated approach significantly boosts system throughput and reduces execution times compared to conventional methods. This study enhances the understanding of the interplay between UAV trajectory planning and task management, offering vital theoretical insights for advancing UAV-assisted MEC systems.

In contrast to traditional mechanism design, the posted pricing mechanism can quickly determine the winning user and ensure the revenue of the seller through a predetermined price. Additionally, the posted pricing mechanism inherently possesses economic properties such as truthfulness and individual rationality. These properties make it an ideal method for solving online task allocation problems for mobile crowdsensing services (MCSs). The challenge in posted pricing mechanism design is being able to find reasonable posted prices under complex MCS task constraints. This paper presents an innovative posted pricing mechanism to solve a general point of interest (POI)-based online MCS task allocation problem. We transform the problem into an integer programming model with the goal of maximizing the total utility of the system while satisfying various constraints. We prove that under any user arrival order, there must exist a posted price structure that can ensure that the total utility of the system is approximately optimal, with an approximation ratio of$1/(d+1)$in the worst case. With the support of theoretical analysis, the posted price calculation can be completed using only a simple gradient descent algorithm. Compared with existing methods, our solution achieves very good results in terms of total utility and the task completion ratio, indicating that it can effectively improve the efficiency and service quality of MCSs.

This paper investigates the performance of Mobile Edge Computing (MEC) systems using the M/M/m queuing model, focusing on how server configurations and queuing rules affect task response times on edge servers. The study considers two types of tasks with different processing requirements: special tasks and general tasks. The study analyzes the response time under two queuing rules: first-come-first-served (FCFS) and head-of-line (HOL) priority. By evaluating the impact of server execution speed and the number of cores, this work provides insights for optimizing task scheduling and resource management in MEC environments, which ultimately improves system performance and reduces latency.

The proliferation of mobile devices and the increasing demand for portable services have led to a surge in computationally intensive and time-sensitive applications, necessitating efficient computation offloading in Mobile Edge Computing. Existing research often overlooks the distinct response time requirements of different applications and their mutual interference during queuing. This paper addresses these gaps by optimizing offloading strategies with multiple response time constraints, focusing on the queuing impact of various tasks. We model edge servers as M/M/1 queuing systems and develop a set of mathematical models. Using algorithms based on Karush-Kuhn-Tucker conditions, we derive optimal offloading strategies to minimize system power consumption. Our approach significantly reduces system power consumption in practical applications, enhancing resource utilization for service providers. Our work introduces a comprehensive and realistic model of task offloading, considering multiple task types, their distinct average response time requirements, and their mutual interference during queuing. This advancement ensures efficient and effective offloading strategies, addressing the complexity of multiple tasks and varying response time constraints.

The explosive growth of terminal devices at the network edge, coupled with advancements in communication technology, poses significant challenges to traditional cloud computing models. Despite Mobile Edge Computing (MEC) mitigating some issues by enabling real-time data processing closer to the source, it faces challenges with linear growth in computational resources insufficient to meet the exponential growth in service demand. Existing research utilizing Software-Defined Networking (SDN), Network Functions Virtualization (NFV), and Service Function Chain (SFC) technologies has made progress but still faces critical issues, such as the inability to respond in real-time to dynamic demands and inefficiencies in resource management strategies. This paper addresses these challenges by optimizing Virtual Network Function (VNF) deployment strategies in MEC environments. We propose PPO-ERA, a novel algorithm leveraging deep reinforcement learning and the Karush-Kuhn-Tucker (KKT) method. This approach provides real-time, adaptive, and dynamic deployment policies for VNFs, significantly improving both the average response delay of tasks and resource utilization. Key contributions include rigorous SFC-based application modeling, dynamic VNF deployment algorithms, elastic resource allocation, uniform state representation, and extensive performance validation. These advancements enhance the adaptability, efficiency, and performance of VNF deployment strategies, addressing critical challenges in dynamic MEC environments.

The rapid expansion of Mobile Edge Computing (MEC), driven by the escalating data volume and the demand for minimal network latency, underscores the need for efficient data processing. To address the growing complexity of neural networks and applications, segmentation into smaller components (e.g., neural network layers, subnetworks, and subtasks) for parallel computation across diverse nodes is common. However, effective data transmission between these segments necessitates optimized task scheduling among edge servers. Many platforms leverage container-based OS-level virtualization to enhance edge computing efficiency, leveraging container image layers to cut storage and transmission costs. However, previous research predominantly emphasizes task scheduling, overlooking runtime environment preparation on edge servers and potential collaboration among edge nodes. This paper introduces an innovative approach that adeptly manages task data and image layer dependencies collaboratively. It formulates an NP-hard problem: minimizing total computation completion time by jointly determining downlink transmission rate allocation, task-offloading strategies, and layer-loading schemes, allowing for thoughtful decoupling and iterative refinement. The Gray Wolf Optimizer and Cellular Automata are introduced for dynamic task scheduling, complemented by a low-complexity algorithm inspired by the Nawas-Enscore-Ham method. For layer downloading, the paper explores a partial-layer loading policy, considering storage constraints, and establishes a full-layer loading strategy with the Peer-to-Peer mechanism, significantly reducing computational complexity. Rigorous experimental results underscore the remarkable efficacy of these approaches in curtailing total computation completion time, positioning them as benchmarks for comparison against alternative solutions.

As an emerging computing paradigm, cloud-edge collaborative computing (CECC) combines computing resources at the back-end and the edge of the network to provide more flexible service delivery, thus striking a good balance between abundant computing resources and high responsiveness. However, mobile devices (MDs) must make strategic offloading decisions in such an environment. Although existing research has made remarkable progress in computation offloading strategies, most works ignore multi-priority settings in complex application scenarios. In this article, we focus on the impact of multi-priority settings and mixed queue disciplines on offloading decisions in CECC. First, we utilize queueing models to characterize all computing nodes in the environment and establish mathematical models to describe the considered scenario. Second, we formulate offloading decisions of the target MD into three multi-variable optimization problems to investigate the cost-performance tradeoff. Third, we propose numerical algorithms based on the Karush-Kuhn-Tucke conditions to address these problems. Finally, we construct numerical examples, a comparative experiment, and a simulation experiment to demonstrate the effectiveness of our methods. Our work provides important insights into the optimization of computation offloading for MDs in complex application scenarios, which can help achieve a better cost-performance tradeoff in CECC.

On-device learning has become a new trend for edge intelligence systems. In this paper, we investigate the on-device in-cremental learning problem, which targets to learn new classes on top of a well-trained model on the device. Incremental learning is known to suffer from catastrophic forgetting, i.e., a model learns new classes at the cost of forgetting the old classes. Inspired by model pruning techniques, we propose a new on-device incremental learning method based on weight freezing. The weight freezing in our framework plays two roles: 1) preserving the knowledge of the old classes; 2) boosting the training procedure. By means of weight freezing, we build up an efficient incremental learning framework which combines knowledge distillation to fine-tune the new model. We conduct extensive experiments on CIFAR100 and compare our method with two existing methods. The experimental results show that our method can achieve higher accuracy after incrementally learning new classes.

The robustness and security of deep neural network (DNN) models have received much attention in recent years. In-depth research on adversarial example generation methods that make DNN models make wrong judgments and decisions will facilitate further research on more comprehensive and practical adversarial defense methods. Most existing adversarial example generation methods focus too much on attack performance and design adversarial noise at the pixel level, resulting in the generated adversarial examples with redundant noise and evident perturbations. In this paper, we try to find the well-designed perturbations at the feature-level and propose a novel deep reversible network-based imperceptible adversarial examples generation method called RIA. Experimental results show that RIA can obtain more natural adversarial examples without losing attack performance and reducing redundant noise based on well-designed feature maps. To the best of our knowledge, in the white-box attack method research, this work is the first attempt to directly add perturbations to feature maps and use an reversible network to generate adversarial examples based on the perturbed feature maps.

As facial image manipulation techniques developed, deep fake detection attracted extensive attentions. Although researchers have made remarkable progresses in deepfake detection recently, which is still suffering from two limitations: a) current detectors achieve high accuracy in the high-quality videos and images, but it is hard to capture local and subtle artifacts in the low-quality and high-compression media; b) few of deep fake detection methods gain satisfying performance under cross-database scenario, because detector overfit to specific color textures producing by same manipulation algorithm. Inspired the above issues, this paper proposes a novel framework fusing local related features and frequency information to mine the forgery patterns. Firstly, we design multi-feature enhancement module, which amplifies implicit local disc repancies and capture spatial correlation from three shallow feature layers and high-level semantic layer guided by attention maps. Secondly, dual frequency decomposition module is proposed for disassembling high-frequency and low-frequency features, the forgery artifacts are exposed after dual cross attention block processing in the frequency spectrum. Features from the two streams are fused to the classification for the final result. Comprehensive experiments demonstrate the superior performance of our proposed approach in the low-quality benchmark database and cross-dataset sce-nario.

As the power density of modern CPUs is gradually increasing, thermal management has become one of the primary concerns for multicore systems, where task scheduling and dynamic voltage/frequency scaling (DVFS) play a pivotal role in effectively managing the system temperature. In this article, we proposeCARTAD, a new reinforcement learning (RL)-based task scheduling and DVFS method for temperature minimization and latency guarantee on multicore systems. The novelty ofCARTADframework is that we exploit the machine learning technique to analyze the applications’ intermediate representations (IRs) generated by a compiler and identify an important feature which is critical for predicting the application’s performance. With the newly explored feature, we construct an RL-based scheduler with the more effective state representation and reward function such that the system temperature can be minimized while guaranteeing applications’ latency. We implement and evaluateCARTADon real platforms in comparison with the state-of-the-art approaches. Experimental results showCARTADcan reduce the maximum temperature by up to 16 °C and the average temperature by up to 10 °C.

Computing resource configuration and site selection of edge servers (ESs) are two critical steps to build up a mobile edge computing (MEC) platform. In this paper, the joint optimization problem of configuration and placement for ES in the MEC environment is investigated. First, we treat each ES as an M/G/m queueing model, and establish mathematical models to characterize the MEC environment, such that the performance and operational expenditures (OPEX) of the system can be calculated analytically. Then, we design a two-stage method and develop a series of algorithms based on bisection algorithm and genetic algorithm (GA) to obtain the optimal configuration scheme and sub-optimal placement scheme (including the deployment quantity) of ESs, with the goal of minimizing OPEX while maintaining system performance at a predetermined level. Finally, we conduct experiments based on a real base station dataset provided by Shanghai Telecom to show the effectiveness of the proposed algorithms. To the best of our knowledge, this work is the first research of the joint optimization problem of configuration and placement for ES in the MEC environment, where the main objective is to increase the cost efficiency.

Abstract Before service providers build up an mobile edge computing (MEC) platform, an important issue that needs to be considered is the configuration of computing resources on edge servers. Since the computing resources on an edge server are limited compared with a cloud server and the service provider's deployment budget is limited, it would be unrealistic to equip all edge servers with abundant computing resources. In addition, the edge servers have different computation demands due to their different geographies. Therefore, this article investigates the problem of server configuration optimization in an MEC environment based on a given computation demand statistics of the selected deployment locations. Our strategy is to treat each edge server as an M/M/m queueing model, and then establish the performance and cost models for the system. Two optimization problems, including cost constrained performance optimization, and performance constrained cost optimization are formulated based on our models and solved by a series of fast numerical algorithms. We also conduct extensive numerical simulation examples to show the effectiveness of the proposed algorithms. MEC service providers can use our strategy to get the appropriate type of processor and obtain the optimal processor number for each edge server to achieve two different goals: (1) deliver the highest‐quality services with a given cost constraint; (2) minimize the investment cost with a service‐quality guarantee. Our research is of great significance for service providers to control the tradeoff between investment cost and service quality.