Research On Computation Offloading And Handoff In Vehicular Edge Computing

In the vehicular cloud computing environment,the cloud servers or other high-performance vehicles are normally used to provide computation offloading services for the resource-constrained vehicles.However,the low transmission rate between the vehicles and the macro base stations may lead to increasing the completion time of the tasks,in the meantime,the massive tasks and trillion-level data would aggravate the load of clouds.How to reduce the completion time of the compute-intensive tasks and to lighten the cloud server load become two emergency problems during the computation offloading.The promotion of edge computation technology offers a new idea for solving these two problems.In the static environment,benefit from closer distances between the edge servers and the vehicles,introducing the edge layer can effectively improve the user experience and release the load.But as the cars driving,the dynamic variety of the communication environment will result in adding the task completion time.As a result,this is the key point of the paper to reduce threats for the computation offloading up to the moving of the automobile.This paper discusses and studies the computational offloading and handoff in the vehicle edge computing environment.Firstly,the task communication,execution and completion time of offloading to edge,or offloading to cloud and local execution are compared in static environment,and the advantages and disadvantages of three implementation schemes of different types of tasks are discussed based on AOL algorithm.Secondly,in view of the vehicle driving state,an offloading algorithm based on handoff strategy is proposed to reduce the negative impact of the increase in the task completion time caused by the computation offloading.Finally,this paper proposes a greedy-based offloading algorithm for path predictability,which further reduces the task completion time.In this study,six performance indicators such as local execution ratio,edge execution ratio,cloud execution ratio,average communication time,average execution time and average completion time were used to evaluate and analyze the offloading performance of the algorithm involved in this paper.The results show that:(1)For the static environment,offloading to edge is reduced by about 42.78%and 73,99%,respectively,compared to the offloading to cloud and executed locally,but the execution time of the cloud is reduced by about 24.47%;(2)For the driving state of the car,the mission-based completion time based on the fastest deceleration algorithm is reduced by about 22.01%compared with the Enumeration-based offloading algorithm;(3)In the path predictable case,when considering the trajectory of the car,the greedy-based offloading algorithm proposed in this paper is reduced by about 14.87%compared with the task-based offloading algorithm.The main innovation of this paper is to propose a computational offloading algorithm based on computational handoff and a link selection strategy based on maximum transmission rate in dynamic environment to reduce the impact of motor motion on computing offloading.Finally,in the case of path predictability,a greedy-based offloading algorithm is designed to further reduce the negative impact of motor motion on computational offloading.