关键词： Task analysis   Delays   Autonomous aerial vehicles   Optimization   Trajectory   Resource management   Energy consumption   Deep reinforcement learning   mobile edge computing   task offloading   unmanned aerial vehicle (UAV)  

摘要： Mobile edge computing (MEC) has emerged as a solution to address the demands of computation-intensive network services by providing computational capabilities at the network edge, thus reducing service delays. Due to the flexible deployment, wide coverage and reliable wireless communication, unmanned aerial vehicles (UAVs) have been employed to assist MEC. This paper investigates the task offloading problem in a UAV-assisted MEC system with collaboration of multiple UAVs, highlighting task priorities and binary offloading mode. We defined the system gain based on energy consumption and task delay. The joint optimization of UAVs' trajectory design, binary offloading decision, computation resources allocation, and communication resources management is formulated as a mixed integer programming problem with the goal of maximizing the long-term average system gain. Considering the discrete-continuous hybrid action space of this problem, we propose a novel deep reinforcement learning (DRL) algorithm based on the latent space to solve it. The evaluation results demonstrate that our proposed algorithm outperforms three state-of-the-art alternative solutions in terms of task delay and system gain.

关键词： Virtual machine placement   Container consolidation   Mobile edge computing   Similarity   Spectral property  

摘要： An optimized container consolidation is a key factor in Mobile Edge Computing. It is a challenging problem as its complexity is NP-hard. In addition to structural properties of a network which reveal nodes features and their connectivity, we believe that the spectral properties of correlation between rows of similarity matrix of the network could be more beneficial in placement algorithm. Most existing literature focus on reducing energy consumption in cloud or edge environments based on structural properties of the given network. However, they undermine the concept of locality which could affect excessive network traffic. Our proposed method utilizes structural and spectral properties of the network to solve the problem. We expected to improve locality factor and link reduction as well as reducing the problem space by utilizing a centroid-based approach and a queue. Our proposed algorithm preserves topology architecture without consuming more energy in comparison to the existing popular algorithms.

关键词： Computational modeling   Task analysis   Artificial intelligence   Sensors   Wireless communication   Data models   Wireless sensor networks   Wireless sensing   collaborative inference   model splitting   mobile edge computing  

摘要： With recent advancements in artificial intelligence (AI), wireless sensing has recently been accepted as an attractive solution to enable accurate detection of human activities by analyzing the radio signal variations of sensor devices (SDs) using a well-trained AI model. However, due to the limited communication and computation resources at SDs, it is impractical to support energy- and delay-sensitive sensing services by solely processing the massive computation workload at local or offloading it to the edge server (ES) for edge inference. To address this problem, we consider in this paper device-edge co-inference in a wireless sensing system where multiple users collaboratively perform a common inference task. In particular, the AI model deployed at each SD can be split into two sequential parts. Each SD executes the former part of AI model at local, and leaves the remaining part computed at the ES. We aim to minimize the energy consumption of SDs subject to a prescribed inference latency requirement. To this end, we formulate a mixed integer non-linear programming (MINLP) to jointly optimize the model splitting point and system resource allocation, where the major difficulty lies in the tight couplings among splitting decisions of collaborative SDs. To solve the problem, we propose an integrated learning and optimization algorithm named LOP, which tackles the combinatorial model splitting by using a deep reinforcement learning (DRL)-based method, and deals with the remaining resource allocation problem via convex optimization. To gain some engineering insights, we study the optimal model splitting design in a practical wireless indoor crowd counting system, where the optimal splitting point exhibits a threshold-based structure related to the user channel gain. Simulation results demonstrate that the proposed LOP algorithm can achieve a near-optimal energy performance with on average 0.8% optimality gap while enjoying a hundredfold reduction in computation delay.

关键词： capsule network   deepfake detection   forgery classification   mobile edge computing   multi-scale feature  

摘要： Face recognition is one of the most important applications of MEC. However, there have been many fake face data that can deceive MEC devices, causing serious problems such as information leakage. Face forgery detection can effectively solve this problem. Current face forgery detection methods have achieved high accuracy. However, most of the methods are researched on the classification of face authenticity. We find that studying multi-classification of forgery methods can not only improve the accuracy of the model to identify fake faces, but also help improve the generalization ability of fake face classification. We argue that multi-scale features and high-frequency features can expose more detailed forgery artifacts. So, we design four modules, which take advantage of the complementarity of RGB features and frequency features, global features and local features. The first module is a residual-guided multi-scale spatial attention module, which uses residuals to guide the RGB feature extractor to extract fake features from a multi-scale perspective. The second module is a multi-scale retinal feature extraction module. The third module is a multi-scale channel attention-guided local frequency statistics module, which extracts local frequency responses using sliding-window DCT. The last module is a capsule network classification module with overall correlation to classify the fused features. The information transfer between the subject capsule and the classification capsule can increase the integrity of the model, making the model converge faster. We conduct experiments on the databases FaceForensics++, DeepfakeDetection, and FakeAVCeleb. Experimental result shows that our method performs well on forgery classification.

关键词： Mobile edge computing   Wireless power technology   Deep reinforcement learning   Markov decision process   Non-orthogonal multiple access  

摘要： Wireless powered mobile edge computing (MEC) has become a vital component of future 6G networks, offering efficient computational capabilities to internet of things (IoT) devices and ensuring a stable energy supply. In this paper, we propose an innovative design for a wireless powered MEC-assisted energy-sensitive IoT systems. The system integrates wireless power technology (WPT), allowing IoT devices to efficiently perform tasks under wireless power supply without consuming their battery energy. Therefore, we formulate a dynamic computational task execution latency minimization (DCTELM) problem to tackle the impacts of unpredictable energy requirements and potential communication latency on resource allocation. We also analyze the differences between time division multiple access and non-orthogonal multiple access communication protocols. Given the complexity and dynamic nature of this problem, we define it as a mixed integer nonlinear programming problem. To address this problem, we introduced an enhanced deep reinforcement learning (DRL) algorithm. First, the DCTELM problem is transformed into a Markov decision process (MDP), which simplifies the structure of the problem. Then, after determining the offloading decision, we reveal the convex relationship between resource allocation and task latency. Furthermore, the action space of MDP is further reduced, and a multihead proximal policy optimization (PPO) algorithm is proposed to deal with the simplified MDP problem. This process reduces the complexity of problem solving, decreases the consumption of computational resources. Simulation results demonstrate that our method outperforms other baselines in terms of computation latency, particularly with the introduction of an exploration strategy during the training phase, which significantly reduces task latency.

关键词： Task analysis   Energy consumption   Switches   Cloud computing   Array signal processing   Minimization   Optimization   Reconfigurable intelligent surface   mobile edge computing   active-passive mode switching   number configuration  

摘要： Mobile edge computing (MEC) has been regarded as a promising technique to support latency-sensitivity and computation-intensive serves. However, the low offloading rate caused by the random channel fading characteristic becomes a major bottleneck in restricting the performance of the MEC. Fortunately, reconfigurable intelligent surface (RIS) can alleviate this problem since it can boost both the spectrum- and energy- efficiency. Different from the existing works adopting either fully active or fully passive RIS, we propose a novel hybrid RIS in which reflecting units can flexibly switch between active and passive modes. To achieve a tradeoff between the latency and energy consumption, an optimization problem is formulated by minimizing the total cost via jointly optimizing the transmission time, the transmit power, the receive beamforming vector, the phase-shift matrix, the mode-switching factor, the amplification factor, the offloading ratio factor, and the computation ability of the user, where the constraints of the maximum energy of users, the maximum power of the RIS, the minimum computation tasks, the transmission time, the offloading ratio factor, the mode-switching factor, the unit moduli of passive units, and the computation ability are taken into account. Considering the complexity of the aforementioned problem, we develop an alternating optimization-based iterative algorithm by combining the successive convex approximation method, the variable substitution, and the singular value decomposition to obtain sub-optimal solutions. Furthermore, in order to gain more insight into the problem, we consider two special cases involving a latency minimization problem and an energy consumption minimization problem, and respectively analyze the tradeoff between the number of active and passive units. Simulation results verify that the proposed algorithm can achieve flexible mode switching and significantly outperforms existing algorithms.

关键词： Ultra reliable low latency communication   Reconfigurable intelligent surfaces   Minimization   Task analysis   Optimization   Computer architecture   Power amplifiers   Active reconfigurable intelligent surface   fully-connected architecture   sub-connected architecture   ultra-reliable and low latency communication   digital twin   alternating optimization   mobile edge computing  

摘要： The substantial power consumption attributed to the active components within fully-connected reconfigurable intelligent surface (RIS) architecture significantly hinders the efficiency and sustainability of DT-enabled MEC networks. To tackle this challenge, we present an innovative sub-connected architecture for active RIS within the digital twin (DT) integrated mobile edge computing (MEC) framework of an Internet-of-Things (IoT) networks, capitalizing on edge intelligence to enhance ultra-reliable and low-latency communication (URLLC) services. The primary aim of our research is to improve uplink data transmission from IoT URLLC user nodes (UNs) to a base station (BS) with the aid of an active RIS, even under an imperfect channel state information (CSI). We have formulated the total end-to-end (e2e) latency minimization problem, which is solved by using an efficient alternating optimization (AO) algorithm. The algorithm breaks down the proposed non-convex problem into five subproblems, namely, beamforming design, caching and offloading policy optimization, joint communication and computation optimization, and joint active RIS phase shift and amplification factor vector optimization. We conducted a thorough analysis of the convergence properties of the proposed AO algorithm, benchmarking its performance against the established Heuristic algorithm. Our simulation results consistently demonstrate the superiority of our proposed DT-assisted optimal phase sub-connected active RIS scheme over various benchmark schemes, taking into account various factors such as the number of RIS elements, power budget constraints, imperfect CSI, edge computing server (ECS) cache capacity, number of IoT UNs, and the number of power amplifiers.

关键词： Vehicular networking   Mobile edge computing   DDPG algorithm   Multi-intelligent body reinforcement learning  

摘要： As a reliable computational model in telematics, Mobile Edge Computing (MEC) is utilized to lighten the load on individual autonomous vehicles. It enables computationally expensive and latency-sensitive jobs to be offloaded from cars with limited processing power to servers at edge nodes. This paper presents the joint optimization problem as a mixed nonlinear integer programming problem. The scheduling of vehicle tasks and the distribution of communication resources are then rationalized using a resource allocation algorithm based on Deep Deterministic Policy Gradient (DDPG) reinforcement learning to arrive at a suboptimal solution based on the resource allocation policies of single-intelligent DDPG and multi-intelligent DDPG. The numerical simulation demonstrates the superior service experience of the proposed joint optimization strategy of distributed task offloading and resource allocation based on multi-intelligent DDPG over other association and channel selection strategies and joint resource allocation based on single-intelligent DDPG.

关键词： Resource management   Optimization   Task analysis   Feature extraction   Collaboration   Network topology   Dynamic scheduling   Intelligent hybrid resource allocation   mobile edge computing   network slicing   recurrent learning   graph reinforcement learning  

摘要： Mobile edge computing (MEC) combining radio access network (RAN) slicing shows tremendous potential in satisfying diverse service level agreement (SLA) demands in future wireless communications. Since the limited computing and transmission capacities, efficient hybrid resource allocation (RA) from the perspective of computing and transmission resources is crucial to maintain a high SLA satisfaction rate (SSR). However, in cooperative multi-node MEC-assisted RAN slicing systems, the complexity of the multi-node cooperation in spatial dimension as well as the contextual correlation of system state in time dimension pose significant challenges to the hybrid RA policy optimization. In this paper, we aim to maximize the SSR for heterogeneous service demands in the cooperative MEC-assisted RAN slicing system by jointly considering the multi-node computing resources cooperation and allocation, the transmission resource blocks (RBs) allocation, and the time-varying dynamicity of the system. To this end, we abstract the system into a weighted undirected topology graph and, then propose a recurrent graph reinforcement learning (RGRL) algorithm to intelligently learn the optimal hybrid RA policy. Therein, the graph neural network (GCN) and the deep deterministic policy gradient (DDPG) are combined to effectively extract spatial features from the equivalent topology graph. Furthermore, a novel time recurrent reinforcement learning framework, namely recurrent actor-critic (RAC), is designed in the proposed RGRL algorithm to cope with the time-varying and contextual network environment. Simulation results in different use case scenarios demonstrate the superiority of the proposed algorithm in terms of the average SSR, the performance stability, and the network complexity.

关键词： incentive mechanism   mobile edge computing   privacy protection   sharding blockchain   smart contract  

摘要： Mobile edge computing (MEC) technology is widely used for real-time and bandwidth-intensive services, but its underlying heterogeneous architecture may lead to a variety of security and privacy issues. Blockchain provides novel solutions for data security and privacy protection in MEC. However, the scalability of traditional blockchain is difficult to meet the requirements of real-time data processing, and the consensus mechanism is not suitable for resource-constrained devices. Moreover, the access control of MEC data needs to be further improved. Given the above problems, a data privacy protection model based on sharding blockchain and access control is designed in this paper. First, a privacy-preserving platform based on a sharding blockchain is designed. Reputation calculation and improved Proof-of-Work (PoW) consensus mechanism are proposed to accommodate resource-constrained edge devices. The incentive mechanism with rewards and punishments is designed to constrain node behavior. A reward allocation algorithm is proposed to encourage nodes to actively contribute to obtaining more rewards. Second, an access control strategy using ciphertext policy attribute-based encryption (CP-ABE) and RSA is designed. A smart contract is deployed to implement the automatic access control function. The InterPlanetary File System is introduced to alleviate the blockchain storage burden. Finally, we analyze the security of the proposed privacy protection model and statistics of the GAS consumed by the access control policy. The experimental results show that the proposed data privacy protection model achieves fine-grained control of access rights, and has higher throughput and security than traditional blockchain.