Research On Task Migration Algorithms In Edge Computing Platform

In the edge computing environment,tasks generated by terminal devices will be encapsulated into specific virtual machines,and the service quality of the whole platform will be improved by flexible deployment and migration of virtual machines in the edge computing platform.This paper studies the dynamic task migration in the edge computing platform.A dynamic task migration system is built to provide an experimental simulation platform for the research of task migration algorithm in edge computing platform.The task manager generates and deploys dynamic tasks in the edge computing environment simulator.The resource predictor predicts the resource usage of the platform in the future,and cooperates with the migration decision-maker to accomplish task migration.In this paper,an efficient task migration(GC-ETM)algorithm based on graph coloring is proposed.On the premise of achieving uniform task allocation,the optimization goal is to reduce energy consumption,communication cost,migration cost and the comprehensive cost of the three.The algorithm is based on the transformation of physical network topology,the preprocessing operation of coloring and resource prediction.The preprocessing process is mainly to realize the real-time recording of server information.The resource prediction is mainly to realize the accurate division of server resource usage by predicting the impact of dynamic changes of tasks on resource requirements in the future.GC-ETM algorithm reduces the communication cost by migrating some tasks on the "overloaded" server;correspondingly,reduces the energy consumption cost by migrating all tasks on the "underloaded" server and shutting them down.When making the migration decision,it will integrate the task selection results and preprocessing results to realize the quick and accurate determination of the optimal migration scheme and reduce the migration cost.In the experiment,the GC-ETM algorithm is compared with BGM-BLA,AVMM and VMCUP-M algorithm in the edge computing environment simulator,which proves that the algorithm proposed in this paper can not only solve the dynamic task migration problem,but also deal with the static task migration problem.At the same time,the algorithm studied in this paper has obvious advantages in the aspects of energy consumption,communication cost,migration cost,average migration cost and comprehensive cost.