Research On Offloading Strategy Of Dividable Tasks In Mobile Edge Computing

As a key technology in 5G,Mobile Edge Computing(MEC)technology “sinks” the traditional cloud computing platform to the edge of the network to provide nearby mobile users with computing and storage services.Due to the limited resources of mobile terminal equipment,it is difficult to run some computationally intensive applications(such as VR/AR,autonomous driving,telemedicine,etc.)on its own.Therefore,mobile terminal equipment can use computing task offloading technology to remove the computationally intensive application part.Unloading to the MEC server,MEC assists the terminal equipment to complete.Compared with the traditional mobile cloud computing(Mobile Cloud Computing,MCC)that uses cloud data centers to provide computing and storage services,MEC has the following advantages: First,MEC is closer to the terminal equipment,effectively reducing the transmission delay between users and MCC;Second,because MEC is located at the edge of the network,it can more quickly perceive changes in the location and environment of surrounding user equipment,so its scheduling will be more flexible;third,MEC is deployed in a distributed manner,avoiding the transmission of massive data to the cloud data center.Effectively alleviate the burden on the core network.Based on the above advantages,MEC has received extensive attention from academia.The current research on the offloading of splittable tasks in MEC generally has the following problems: ignore the dependencies between subtasks after task splitting,or only consider the order dependencies of tasks;simplify the processor model of the MEC server,assuming the server The processing power is infinite,so that the problem of heterogeneous computing and storage capabilities of multiple MEC servers is ignored;most studies only make decisions from the perspective of mobile devices,such as controlling the transmission power of the device and the CPU frequency,but rarely from the characteristics of the computing tasks To formulate an uninstallation strategy from the perspective of the United States;ignoring the urgency of the application to be processed,it is essential to ensure that the task is completed within the tolerance time for tasks involving safety in the Internet of Vehicles or telemedicine fields.Therefore,in response to the above problems,this paper studies the multi-edge server collaboration scenario combining dense cell network technology and mobile edge computing technology and the decision-making problem of divisible task offloading in the D2 D edge computing scenario.The specific research is mainly as follows:(1)Consider the collaborative scenario of multiple MEC servers in a dense cell,and combine dense cell technology with MEC technology to provide users with higher system network capacity and computing capabilities.This paper is based on Directed Acyclic Graph(DAG)to model computationally intensive applications.The application can be divided into multiple subtasks with specific dependencies.In order to improve the quality of user experience and reduce the communication and calculation costs of mobile users,a system cost optimization problem with application completion time constraints is formulated.The problem proposed is NP-hard.Therefore,a Task Allocation Algorithm based on Task Clustering(TAA-TC)is proposed to solve the optimization problem.The algorithm is divided into three steps: first,the DAG task model is used to cluster adjacent subtasks in the vertical direction;second,the priority of the clustered task clusters is determined,and the maximum tolerable delay of each task cluster is calculated;and finally,Through the priority between task clusters,the maximum tolerable delay of task clusters,and the parallelism between task clusters in the horizontal direction,task allocation is performed.This article first discusses task allocation to tree structure applications,and then discusses the task allocation problem of general-purpose applications,simplifies general-purpose applications into tree-structure task graphs,and uses the proposed algorithm to allocate tasks.Finally,the effectiveness of the algorithm is verified through experimental simulation.(2)Under the trend of large-scale growth of Io T devices,this article hopes to make full use of the computing resources of these mobile devices and make full use of the advantages of deviceto-device(Device-to-Device,D2D)communication technology,and propose a D2 D mobile edge computing task offload architecture,Mobile devices can use its direct end-to-end communication method to share computing resources with each other in real time.Among them,the locally idle wireless device(Wireless Device,WD)serves as an edge computing node near the local user and provides computing resources for the local busy user equipment.This paper firstly models multiple computationally intensive applications based on Directed Acyclic Graph(DAG).In this model,each application contains multiple subtasks,and there are strict dependencies between subtasks..Second,in view of the particularity of the D2 D edge computing system(the edge computing node is composed of mobile smart devices with limited battery capacity),a new indicator of the average energy consumption ratio of the system is proposed to evaluate the energy consumption of each mobile device in the system.Based on the system architecture,task modeling and new evaluation indicators,the optimization problem of minimizing the average energy consumption rate of the system under the condition of meeting the application completion time limit is formulated.Considering the NP-hard nature of the problem proposed,this paper transforms the optimization problem into a task scheduling problem,and considers the urgency of the application to be processed,and proposes a heuristic task scheduling algorithm based on task priority(Heuristic Task Scheduling Algorithm based on Task Priority,HTSA-TP),to solve the optimization problem.Finally,the effectiveness of the algorithm is verified through experimental simulation.