Research On Computation Offloading Strategy In D2D-assisted Mobile Edge Computing Networks

The explosive growth of mobile data traffic and the vigorous development of computation-intensive applications have brought great challenges to the data processing of mobile communication networks.To address this problem,mobile edge computing(MEC)deploys services and functions originally located in the cloud server to the edge of the network to conduct short-distance computation offloading from users to MEC server,so as to meet the key requirements of low latency and high computing power in computation-intensive applications.However,due to the limited resources of MEC server,when users' requests exceed its service capacity,MEC offloading is difficult to meet the requirements of all users.At this time,the multi-user collaborative computing mode based on Device-to-Device(D2D)communication and the cloud computing mode can be introduced to perform computation offloading jointly with MEC.In the network where MEC offloading,D2D offloading and cloud offloading modes coexist,the key of the research is to design an effective computation offloading strategy to achieve the goal of increasing users'revenue and reducing the offloading cost.The main work of this thesis is as follows:Under the coverage of a single MEC server,considering the situation that MEC may be overloaded when there are many offloading requests from users,D2D offloading is introduced to share the pressure of MEC,and we study the task offloading and resource allocation strategy with D2D-assisted MEC.Users can divide their tasks and offload portions of the tasks to a MEC server or a set of service users for processing.Firstly,in order to motivate users to participate in the computation offloading,we design a user influence evaluation mechanism based on the historical offloading relationship and define the users' social revenue function.Secondly,we formulate the optimization problem of maximizing the social revenue with the joint consideration of offloading modes selection,task segmentation,MEC computation resource allocation and D2D user selection.Then,in order to solve this problem,the offloading modes selection is modeled as a potential game and the existence of a Nash equilibrium is proved.On this basis,the MEC computation resource allocation problem and the D2D user selection problem are solved by using Lagrange multiplier method and greedy strategy respectively.Finally,a joint task offloading and resource allocation algorithm is proposed based on the potential game to obtain the suboptimal solution of the original problem.The simulation results show that the proposed scheme can effectively increase the users' revenue and reduce the task execution delay compared with the benchmark schemes.For the tasks associated with specific application services,such as cloud games,target identification,etc.,the computing nodes need to cache the corresponding services when performing these tasks.Under the coverage of the cloud and multiple MEC servers,considering the interaction between caching and offloading,we study the joint optimization of caching and offloading with D2D-assisted cloud-edge collaboration.Firstly,we consider the scheme that three types of offloading auxiliary nodes including MEC servers,the cloud computing center and user equipment coexist,and model the service caching,task offloading and users' social relationship.Secondly,we define the offloading cost as the weighted sum of task execution delay and energy consumption,and formulate the joint optimization problem of service caching and task offloading with the objective of minimizing the offloading cost.Then,the problems of computation resource allocation,power allocation and task segmentation in the MEC offloading,cloud offloading and D2D offloading modes are solved respectively.On this basis,an auxiliary node selection algorithm is used to match the requesting users and auxiliary nodes.Finally,a joint decision-making algorithm of caching and offloading based on discrete particle swarm optimization is designed.This algorithm obtains the suboptimal solution of the original problem by alternating iterative optimization of service caching and task offloading.The simulation results show that the proposed scheme can effectively reduce the total offloading cost in the network,and its caching and offloading decisions both show the superiority compared with the benchmark schemes.