Research On Computation Offloading In Cooperative Vehicle Infrastructure System

The realization of traditional cloud computing needs to go through the core network.The frequent interaction process between the On Board Unit and the remote cloud will increase the network burden,resulting in the critical path blockage of the core network,which cannot meet the increasingly high demand for low delay of the Internet of Vehicles.Mobile edge computing is proposed as a promising solution,which not only realizes the real-time service of vehicle terminal,but also greatly reduces the battery consumption of vehicle terminal.However,different from remote cloud,edge cloud is realized by distributed small-scale server due to the limitation of external environment of base station and its own deployment cost.And with the advent of the Internet of Everything era,a large number of fragmented intelligent terminals offload data to the edge cloud for computing,which exceeds the computing capacity of the latter.Therefore,the performance of traditional time-consuming and energy-consuming network methods in the Internet of Vehicles will be greatly reduced.In this paper,under the research background of the vehicle platoon assisted by the edge cloud server,the computing resources allocated by the edge cloud server for the vehicle formation and the idle computing resources of other members of the fleet,as well as the distributed computing among computing units,parallel computing within computing units and the delay constraints of computing tasks are considered.The computation offloading problem is studied as follows:(1)The framework of computation offloading in Cooperative Vehicle Infrastructure System is presented,the functions of each component are introduced,and the process of computation offloading and resource allocation is analyzed.(2)Considering the limited heterogeneous network resources of the Internet of Vehicles,the limited computing resources allocated by the edge cloud server for the vehicle platoon,the limited computing resources of the On Board Unit of the member vehicles and the different delay constraints of the computing tasks,a computation offloading model was established with the optimization objective of reducing the total energy consumption of the fleet.(3)The task offloading decision and resource allocation strategy jointly optimize the task offloading and resource allocation schemes.The former adopts the Generalized Benders Decomposition and the extended Adaptive Large Neighborhood Search algorithm respectively,while the latter adopts the method that makes the computing task complete within the delay constraint.(4)The computation offloading simulation environment was built based on Edge Cloud Sim,and the energy consumption optimization comparison experiment of the two task offloading decision algorithms in this paper was conducted,and the experimental results shows the effectiveness of the extended Adaptive Large Neighborhood Search algorithm in optimizing the total energy consumption of the fleet in the decision process of computation offloading.