Research On Dynamic Available Computing Resources Based Computation Offloading Handoff Strategy

In vehicle cloud computing,computation tasks are always offloaded to the cloud servers.However,when massive computation tasks are all offloaded to the cloud would lead to overload and reduce the quality of users’ experience.Therefor,How to reduce service delay and alleviate the congestion of core network have become the main researches in the field of computing offloading of vehicles.Vehicle edge computing effectively solves the above problems by deploying edge servers on the RSU which close to vehicle users to provide localized cloud services with high-bandwidth and lowlatency.While the mobility of the vehicle and the concurrent executing feature of the offloaded tasks make the offloading scenarios changed dynamically.Therefore,how to effectively deal with the impact of the dynamic offloading scenarios on the service performance is the main direction of this paper.In this paper,the experiment simulations are carried out to investigate the effect of dynamic available computing resources on the offloading performances in the vehicular edge computing environment.Firstly,this paper has proved the impact of introducting of the edge computing layer on the offload performances in the static offloading scenarios,then,the selection strategy of uploading to the target RSU with a fixed path is proposed for mobility.Finally,for the scenarios of dynamic change of available computing resources,the handoff algorithm based on dynamic available computing resources is proposed.Experiments prove that the proposed algorithm can effectively deal with the dynamic change of the computing resources.These experiments are based on performance indicators such as average task uploading time,average task communication time,average task execution time,and average task completion time.The experimental results show that:(1)In the static scenarios,the introduction of the edge layer has reduced the average task completion time by 67.16% and 48.14% compares with local layer and cloud layer;(2)For the dynamic offloading scenarios,the selection strategy based on fixed path compares with the random selection strategy and the nearby selection strategy has reduced by 33.89% and 20.78% on the average task uploading time.(3)For the impact of the dynamic change of the available computing resources of agent resources on the computing offloading,the algorithm proposed in this paper is reduced by 21.94% and 9.86% respectively on the average task completion time compares with the Minimum Completion Time(MCT)algorithm and the Handoff based Minimum Execution Time(METH)algorithm.Experiment results have proved the effectiveness of the algorithm proposed in this paper.