Research On Task Offloading And Resource Deployment For Mobile Edge Computing

With the explosive growth of Internet-of-Things and tremendous Io T applications,the number and heterogeneity of Io T user demands on computational and communication resources have been growing exponentially,which have pushed the horizon of a novel computing paradigm-mobile edge computing(MEC).In MEC,a number of edge servers are deployed close to users,and the computation-intensive tasks from end users are uploaded to and processed in the network of edge servers.After that,the results are returned to the end user.In this way,the completion time of services and the energy consumption of end users can be reduced.During this process,task offloading,edge resource allocation and task allocation are three important issues in edge computing and have attracted continuous research attention in recent years.The key issue of task offloading is where to upload each task so as to better reap the benefits of available resource.However,existing research works on task offloading often overlook the unique task topologies and schedules from the Io T devices,which lead to degraded offloading performance.The critical issue of edge resource allocation is how to optimize the deployment location of each service function so as to enhance edge resource utilization.However,the works focus on edge resource allocation often overlook the geographic distribution of users,the dynamic of wireless condition,and the coverage of the edge server.As a result,the edge computing and communication resources cannot be fully utilized.Task allocation is the key problem in D2D edge networks.However,the current studies focus on task allocation often overlook the uncertain processing time of the tasks,which will lead to a decline in overall processing efficiency.To fill the gap,this dissertation focus on these unsolved problems of task offloading,edge resources management and task allocation.1.Task offloading decision.The existing offloading schemes often overlook the task topologies,which results in degraded performance.To solve this problem,this dissertation investigates the problem of fine-grained task offloading in edge computing for low-power Io T systems.By explicitly considering 1)the topology/schedules of the Io T tasks,2)the heterogeneous resources on edge servers and,3)the wireless interference in the multi-access edge networks,this dissertation proposes a lightweight yet efficient offloading scheme for multi-user edge systems,which offloads the most appropriate Io T tasks/subtasks to edge servers such that the expected execution time is minimized.This work has been published in IEEE SECON 2019 and IEEE Internet-of-Things Journal.2.Edge resource allocation.The current edge resource allocation strategy often overlooks the time-varying of geographical distribution and the wireless dynamics.As a result,the edge resource cannot be fully utilized.Based on this observation,this dissertation proposes a deep reinforcement learning(DRL)based approach for edge resource deployment,which has the following two salient features.First,we employ DRL to estimate the underlying wireless features affected by user mobility,which has a direct impact on the task performance in the edge-Io T systems.Second,we implicitly utilize the multi-access opportunity to deploy the NFs,considering the estimated features and user requests.This work has been published in IEEE Internet-of-Things and IEEE Transactions on Computational Social Systems.Considering the limited communication range and high transmission rate of next generation communication technology,the current edge computing frameworks are no longer appropriate for edge computing.To solve this problem,we propose a mobile resourcesharing framework that employs mobile edge servers to provide a cost-effective deployment of edge computing,which enables edge resource sharing for massive Io T devices.MESF exploits the synchronization between requests receiving and results returning.As a result,the server moving and task processing could be paralleled and the overall task efficiency can be improved.3.Task allocation in D2D edge networks.Most existing works on task offloading assume the task processing time is known or can be accurately estimated.However,the processing time is often uncertain until it is finished.Moreover,the same task can have largely different execution time under different scenarios,which leads to inaccurate offloading decisions and degraded performance.To address the problem,this dissertation proposes a game-based probabilistic task offloading scheme with uncertain processing time in D2D edge networks.First,we characterize the uncertainty of task processing time using a probabilistic model.Second,we incorporate the proposed probabilistic model into an offloading decision game.We also analyze the structural property of the game and prove it can reach a Nash equilibrium.