Research On Resource Allocation Of Federated Cache In Mobile Edge Computing

Mobile Edge Computing(MEC)is one of the most promising technologies in the next generation of wireless communication systems,which effectively solves the problem of high latency response requirements and insufficient computing power of users by deploying servers distributed at the edge of the network to sink computing closer to users.However,with the increasingly stringent requirements of Internet of Things(Io T)devices on latency and battery life,the development of MEC technology has also ushered in many challenges,among which resource allocation and computation offloading research has been widely valued by academia.Since edge nodes have limited compute and storage resources,we need to allocate these resources appropriately in order to utilize them as efficiently as possible.How to formulate flexible computation offloading and resource allocation strategies according to the existing complex network environment and dynamically changing resources of various dimensions to fully tap the performance gain of MEC system in terms of delay and energy consumption is of crucial significance.In view of the above challenges,the main research work of this paper is as follows:(1)In the mobile edge computing scenario of multi-user and multi-task,the computation offloading and resource allocation strategy based on task queue is studied considering the channel conditions and the situation that the task arrival is unknown.Firstly,the network model and task queue model of MEC system are established,and the optimization goal of minimizing the long-term average energy consumption of the system is proposed under the conditions of stable task queue and delay limitation of the system model.Then,the Lyapunov optimization method is applied to decouple the time coupling of the task queue,decouple the multi-stage mixed integer nonlinear programming(MINLP)problem into a common optimization problem,and use the deep reinforcement learning(DRL)method to solve the decoupled single-slot resource allocation strategy.Finally,simulation experiments show that the proposed algorithm can stabilize the task queue and reduce the system energy consumption under the arrival of heavy task data,and improve the task execution success rate by 10%～20%.(2)The computation offloading,resource allocation and edge cache joint optimization problems of cache-assisted multi-user MEC system are studied.Firstly,for the allocation of computing,communication and cache resources,the joint optimization model of computing model,communication model and cache model is established.It focuses on the caching mechanism of the task,and divides the cache strategy into three steps: cache value calculation,cache decision and cache replacement.The popularity of cached content is measured according to the number of requests to be cached and the proportion of calculation delay,so that tasks with high cache value are cached in the MEC server to reduce duplication during unloading.At the same time,considering the limited storage space,the cache content is periodically replaced.Then,according to the joint optimization model,a balanced optimization problem model that minimizes delay and energy consumption is proposed.Finally,a DRL-based joint optimization strategy for multi-dimensional resource allocation based on DRL is proposed,and evaluated by simulation experiments.Experiments show that the proposed algorithm can better meet the requirements of low energy consumption and low latency of task execution.