Research On Task Offloading Methods For Users With Different Needs In Mobile Edge Computing

At present,a lot of research on task offloading in mobile edge computing have been accumulated academically,but there are still some shortcomings.On the one hand,the current research works mainly focus on minimizing the average delay or minimizing the overall delay as the optimization goal,ignoring the differentiated delay requirements of different tasks,which will result in many tasks that cannot be completed in accordance with the delay constraints.On the other hand,for the problem of user quality of service(QoS)degradation caused by mobility,current research only considers how to improve QoS through migration or dynamic offloading,ignoring the interdependence between the two and the common impact on QoS,and they are also mainly to minimize the overall delay which makes it hard to achieve better QoS.In view of the shortcomings of the above research,this article will study the task offloading method for users' differentiated demands under the mobile edge computing environment,consider the influence of users'differentiated delay requirements and user mobility,and try to make the tasks complete on time,thereby improving users QoS.This article will analyze the application of deep reinforcement learning in task offloading from the two dimensions of user QoS perception and user mobility perception,and discuss its effectiveness progressively,as follows:(1)Research on user QoS-aware task offloading methodIn view of the users' differentiated task delay requirements,highlighting the impact of the delay requirements of each task on its offloading decision,this article will study the user QoS-aware task offloading problem,and propose a multi-user task offloading decision algorithm based on deep reinforcement learning,which transforms the current environmental information and task completion status into the basic elements of reinforcement learning,and guide the agent to achieve the goal of maximizing the number of tasks completed on time.In the simulation experiment,the algorithm outperforms the benchmark experiment in terms of the number of tasks completed on time.(2)Research on user mobility-aware task offloading and migration methodUser movement may lead to task migration or cross-area services,resulting in increased delay and severely affecting user QoS.On the basis of the previous research point,this article will study the user mobility-aware task offloading method to prevent QoS degradation,propose a multi-user task offloading and migration decision algorithm based on deep reinforcement learning to meet users' differentiated delay requirements in the mobile edge computing.This algorithm transforms current environment information and task completion status into basic elements of reinforcement learning,and the agent is guided to make decisions about whether to offload or migrate tasks and where to migrate,to achieve the goal of maximizing the number of tasks completed.The experiment uses real human movement trajectory data,which proves the effectiveness of the algorithm in the indicator of the number of tasks completed.Through the two perspectives of user QoS awareness and user mobility awareness,this article will specifically elaborate on task offloading and migration issues and the process of modeling using deep reinforcement learning,and propose decision-making algorithms based on deep reinforcement learning,which will provide some reference for future research on task offloading in mobile edge computing.