Research On Edge Computing Resource Allocation Method Based On Multiple Latency Objectives

With the widespread application of various terminal mobile devices and the continuous emergence of computationally intensive applications such as video analysis,virtual reality,speech recognition,etc.,the mobile terminal generates a large amount of data,which has a high requirements on storage and calculation.It highlights the problem for insufficient storage and computing power on the mobile terminals.For that reasons,Mobile Edge Computing(MEC)technology was proposed to solve these problems.The MEC technology deploys MEC servers with strong storage and computing capabilities in network edge environments such as wireless access nodes,allowing computing tasks of users are offloaded to the adjacent network edge for execution,which can effectively avoid the rapid energy consumption of mobile devices to perform tasks,and significantly reduce service delay compared to cloud computing.However,in the MEC system,the computing resources of the MEC server are limited.Therefore,it is of great significance to rationally schedule and allocate resources for tasks that are offloaded from mobile devices to improve system resource utilization and reduce system delay.This paper studies the resource allocation and task scheduling problems in a typical multi-user MEC network scenario.The specific contents are summarized as follows:(1)For the problem of resource allocation in the multi-user MEC system scenario where the number of tasks is less than the maximum number of tasks assigned and executed by the MEC server at the same time,the situation of sufficient and insufficient resources is considered separately.For the former case,we set two optimization objectives based on the actual situation,namely,minimizing the maximum delay and minimizing the total delay.And based on these two delay optimization goals,construct a resource allocation problem model under the maximum tolerable delay constraint.After determining the transmission channel bandwidth allocation,the problem is simplified to the computational resource allocation problem.Then,based on the preallocation and KKT conditions,resource allocation algorithms for these two delay optimization targets are proposed.For the latter case,build a resource allocation problem model based on minimizing the delay more than the limit,and then convert the problem into a more solvable problem by analyzing the optimal solution conditions,use a variant of the algorithm proposed for former case to solve it.The resource allocation algorithm for this problem is obtained.Simulation results confirm the effectiveness of the proposed resource allocation algorithm.(2)For the problem of resource allocation and task scheduling in the multi-user MEC system scenario where the number of tasks is more than the maximum number of tasks assigned and executed by the MEC server at the same time.Combined with the actual situation,two kinds of optimization goals,namely,minimizing the sum of the maximum delay and the time delay exceeding the limit,minimizing the sum of the total delay and the time delay exceeding the limit is determined.And the resource allocation and task scheduling models are modeled based on these two optimization goals.Then,a round-based resource redistribution strategy is proposed for tasks scheduling and resource allocation.The simulation results confirm that the proposed algorithm can effectively reduce the system delay,and at the same time make the proportion of tasks completed within the delay limit increased.