关键词： Computational modeling   Servers   Routing   Training   Network architecture   Machine learning   Stars   Mobile edge computing   federated learning   in-network computation   large-scale distributed learning  

摘要： Most conventional Federated Learning (FL) models are using a star network topology where all users aggregate their local models at a single server (e.g., a cloud server). That causes significant overhead in terms of both communications and computing at the server, delaying the training process, especially for large scale FL systems with straggling nodes. This article proposes a novel edge network architecture that enables decentralizing the model aggregation process at the server, thereby significantly reducing the training delay for the whole FL network. Specifically, we design a highly-effective in-network computation framework (INC) consisting of a user scheduling mechanism, an in-network aggregation process (INA) which is designed for both primal- and primal-dual methods in distributed machine learning problems, and a network routing algorithm with theoretical performance bounds. The in-network aggregation process, which is implemented at edge nodes and cloud node, can adapt two typical methods to allow edge networks to effectively solve the distributed machine learning problems. Under the proposed INA, we then formulate a joint routing and resource optimization problem, aiming to minimize the aggregation latency. The problem turns out to be NP-hard, and thus we propose a polynomial time routing algorithm which can achieve near optimal performance with a theoretical bound. Simulation results showed that the proposed algorithm can achieve more than 99% of the optimal solution and reduce the FL training latency, up to 5.6 times w.r.t other baselines. The proposed INC framework can not only help reduce the FL training latency but also significantly decrease cloud's traffic and computing overhead. By embedding the computing/aggregation tasks at the edge nodes and leveraging the multi-layer edge-network architecture, the INC framework can liberate FL from the star topology to enable large-scale FL.

关键词： Mobile edge computing   Cache pollution attack   Flash crowd  

摘要： Through caching popular contents at the network edge,wireless edge caching can greatly reduce both the content request latency at mobile devices and the traffic burden at the core network.However,popularity-based caching strategies are vulnerable to Cache Pollution Attacks(CPAs)due to the weak security protection at both edge nodes and mobile devices.In CPAs,through initiating a large number of requests for unpopular contents,malicious users can pollute the edge caching space and degrade the caching efficiency.This paper firstly integrates the dynamic nature of content request and mobile devices into the edge caching framework,and introduces an eavesdroppingbased CPA strategy.Then,an edge caching mechanism,which contains a Request Pattern Change-based Cache Pollution Detection(RPC2PD)algorithm and an Attack-aware Cache Defense(ACD)algorithm,is proposed to defend against CPAs.Simulation results show that the proposed mechanism could effectively suppress the effects of CPAs on the caching performance and improve the cache hit ratio.

关键词： Mobile edge computing   Computation offloading   Unmanned aerial vehicle   Digital twin   Deep reinforcement learning   Federated learning   RESOURCE-ALLOCATION   ENERGY-EFFICIENT   COMMUNICATION   OPTIMIZATION  

摘要： In the industrial Internet, Mobile Edge Computing (MEC) can provide the ability to transfer a large number of delay-sensitive and compute-intensive tasks to MEC servers, thus improving Quality of Service (QoS). Considering Unmanned Aerial Vehicles (UAVs) have the advantages of wide communication coverage and low deployment cost, UAVs have great potential to be employed as aerial base stations to provide computation resources for Intelligent Mobile Devices (IMDs). Due to the limited computation resources and energy of IMDs, we designed a multi-UAV-assisted MEC system in ground cells. To minimize the weighted sum of task completion delay and energy consumption, and ensure the QoS requirements of IMDs, we jointly consider the dynamic channel state, renewable energy utilization, UAVs trajectory, and tasks offloading ratio. To solve the non-convexity problem of complex high-dimensional states, we propose a model-free Deep Reinforcement Learning (DRL) offloading scheme based on the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm. Moreover, we adopt Federated Learning (FL) to train DRL models to enhance the robustness of the model and the security of IMDs data. Meanwhile, the real environment is modeled as Digital Twin (DT) to monitor network changes and train the local DRL model, and the central cloud server can obtain the local model in real-time to aggregate the global model. Extensive experimental numerical results show that the proposed algorithm improves the system energy efficiency and reduces task completion delay.

关键词： Mobile edge computing   Green communications   Mixed-integer programming   Resource allocation  

摘要： We investigate the green resource allocation to minimize the energy consumption of the users in mobile edge computing systems,where task offloading decisions,transmit power,and computation resource allocation are jointly optimized.The considered energy consumption minimization problem is a non-convex mixed-integer nonlinear programming problem,which is challenging to solve.Therefore,we develop a joint search and Successive Convex Approximation(SCA)scheme to optimize the non-integer variables and integer variables in the inner loop and outer loop,respectively.Specifically,in the inner loop,we solve the optimization problem with fixed task offloading decisions.Due to the non-convex objective function and constraints,this optimization problem is still non-convex,and thus we employ the SCA method to obtain a solution satisfying the Karush-Kuhn-Tucker conditions.In the outer loop,we optimize the offloading decisions through exhaustive search.However,the computational complexity of the exhaustive search method is greatly high.To reduce the complexity,a heuristic scheme is proposed to obtain a sub-optimal solution.Simulation results demonstrate the effectiveness of the developed schemes.

关键词： Virtual reality   Mobile edge computing   Multimedia communication   Distributed algorithms   EDGE   DELIVERY  

摘要： Virtual Reality (VR) is becoming an important use case for 5G wireless networks, and Mobile Edge Computing (MEC) servers are being explored as a way to reduce VR video latency. To address the limited cache space of a single MEC server, this paper proposes dividing geographically close MEC servers into collaborative domains. A new VR video transmission architecture is designed after analyzing video compression mechanisms.The proposed architecture splits the VR video into equal-sized tile files to simplify the cache problem and improve caching efficiency. However, the large number of different tile files presents a challenge. To address this, the paper proposes an optimized k-shortest paths (OKSP) algorithm with a time complexity of O((K & BULL; M+N)& BULL; M & BULL;log N), where K is the number of tiles that all M MEC servers can cache in the collaboration domain and N is the number of tile files. For extremely large-scale data cases, a greedy-based approximation algorithm is also proposed. The numerical results demonstrate the OKSP algorithm's excellent performance in solving large-scale data, outperforming other caching algorithms in experiments.

关键词： Mobile edge computing   Cloudlet deployment   Graph cut   Approximate kernel optimization   SERVICE PLACEMENT   EDGE  

摘要： Mobile edge computing provides a low-latency, high-bandwidth cloud computing environment for resource-constrained mobile devices by allowing mobile devices to offload tasks, but user task migration causes greater transmission delays. Cloudlets, a new component of mobile edge computing, can perform tasks offloaded by mobile users nearby to reduce the access latency and meet users' requirements for system response time. However, deploying cloudlets in large-scale wireless metropolitan area networks (WMANs) to improve the service quality of mobile applications is currently still difficult. To resolve this issue, we design a cloudlet deployment model based on approximate graph cut, which abstracts the wireless communication network into an undirected weighted graph, divides the graph according to the access point location attributes, and minimizes the user access delay of subgraphs to obtain optimal network area segmentation and cloudlet deployment locations. We also develop an efficient kernel method to optimize the objective function of graph cuts. The simulation experimental results demonstrate that our model has low time and space complexity; thus, it is suitable for large-scale cloudlet deployment and has valuable application prospects.

关键词： Computational capacity   disaster management   mobile edge computing   offloading   unmanned aerial vehicles   RESOURCE-ALLOCATION   ENERGY   BLOCKCHAIN   IOT  

摘要： Unmanned aerial vehicles (UAVs) are considered for disaster management to provide better coverage or help first responders provide advanced services. Furthermore, UAVs combined with mobile edge computing (MEC) capabilities, have recently emerged as a support for cloud computing by deploying UAV-assisted MEC at the network edge to reduce the latency and load of cloud computing data centers. We investigate the problem of maximizing the utility function by jointly optimizing the computing power, user associations, performance, duration, and location of user equipments (UEs) in UAV-assisted MEC networks in a disaster scenario. We considered several constraints including delay, quality of service, and coverage. The optimization problem is a mixed-integer non-linear programming problem. We propose a multi-stage offloading algorithm based on a learning algorithm and an interior-point method to obtain a viable solution. The simulation results obtained demonstrate the effectiveness of our proposed algorithm compared to existing schemes.

关键词： Mobile edge computing   Wireless energy transfer   Intelligent reflecting surfaces   Unmanned aerial vehicle   RESOURCE-ALLOCATION   BEAMFORMING DESIGN   RATE MAXIMIZATION   OPTIMIZATION   NETWORKS   TASK   NOMA  

摘要： Mobile edge computing (MEC) and wireless energy transfer are two important paradigms to improve the computation bits and provide more cost-effective energy for low-power internet of things (IoT) devices. However, the performance of energy transfer and task offloading is significantly affected by propagation loss, and hence the links between IoT devices and transmitters are frequently blocked. To address this issue, in this paper, an intelligent reflecting surfaces (IRS) system is utilized to enhance the energy harvesting and task offloading performances in a multi-unmanned aerial vehicle (UAV)-assisted MEC system. Also, a power beacon (PB) is deployed in the coverage area to provide suitable and stable energy for UAVs via a laser charging system. The sum computation bits of all IoT devices are maximized by jointly optimizing transmit power, time allocation, the phase shift in the energy transfer stage, the phase shift in the task offloading stage, CPU frequency of IoT devices, and UAVs' trajectory. An alternate search method (ASM) is proposed to tackle the non-convexity of the formulated problem in which the closed-form expressions of transmit power, time allocation, the phase shift in the task offloading stage, and CPU frequency are derived. Also, the semidefinite relaxation (SDR) approach and successive convex approximation (SCA) methods are exploited to obtain the phase shifts in the energy transfer stage and UAVs' trajectory via CVX optimizer, respectively. The numerical evaluations confirm the performance gain of the proposed IRS-assisted multi-UAV MEC system compared to existing benchmark designs.& COPY; 2023 Elsevier Inc. All rights reserved.

关键词： Smart factory   Mobile edge computing   Cloud-edge-end collaboration   Emergency task offloading   OPTIMIZATION  

摘要： The realization of smart factories is becoming increasingly possible due to the rapid development of edge computing technology and 5G communication technology. Smart devices in smart factories can perform tasks such as environmental monitoring, production scheduling, and autonomous mobility. These tasks cannot be completed on time depending only on the computing power of smart devices. Therefore, some tasks need to be performed on edge servers and cloud servers. However, events such as natural disasters, network attacks, and hardware failures will cause server computing resource failures. These threats bring hidden dangers to the normal operation of smart factories. This paper proposes an emergency offloading strategy based on cloud-edge-end collaboration for smart factories. In this strategy, the total task execution delay and the key task execution delay constitute an objective function, which is solved by a Fast Chemical Reaction Optimization (Fast-CRO) algorithm. The algorithm follows the principle of prioritizing the offloading of key tasks in emergency situations, and it quickly makes emergency offloading decisions for the system. Simulation results show that this strategy outperforms the other five baseline algorithms and existing priority based CRO algorithms.

关键词： Mobile edge computing   Unmanned aerial vehicle   Energy harvesting   Deep reinforcement learning   WIRELESS   MAXIMIZATION  

摘要： In order to meet the double-sided challenges brought by the shortage of computation resources and energy of users, we investigate in this paper the optimization of energy efficiency (EE) in an unmanned aerial vehicle (UAV)-assisted wireless network, where UAV is functioned as a flying energy station and edge server to provide charging and computing services for ground users. We aim to maximize the average EE of the mobile edge computing network by the joint design of user transmit power, user computing frequency, UAV transmit power, bandwidth allocation, and UAV trajectory planning under strict energy and power constraints. In order to solve such challenging problem, we first elaborately construct a Markov decision process to model task offloading and resource allocation by learning from past experiences. Then, an average EE maximization method relying on deep reinforcement learning (DRL) is designed to efficiently adjust task offloading policy, where the policy of agent can be gradually improved by interacting with the environment and collecting the experience for learning. Finally, the EE-maximization proximal policy optimization (EE-PPO) algorithm is proposed to train the DRL agent and thereby solve this optimization problem. Numerical results are given to indicate that the proposed EE-PPO method has the properties of both fast convergence and well performance.