Research On Task Offloading Method Of Internet Of Vehicles Based On Mobile Edge Computing

With the arrival of 5G era,mobile edge computing,as a very important part of5 G architecture,has a very bright future.It can support many innovative services and applications requiring ultra-low latency.As a new technology,Internet of Vehicles（Io V）needs high bandwidth and low delay.Therefore,the introduction of mobile edge computing into the Io V can solve the problems of energy consumption and delay in the Io V.In this paper,the problems of vehicle computing task offloading and vehicle service migration in the mobile edge computing environment of the Internet of vehicles are analyzed respectively,and the offloading scheduling and service migration methods are proposed through the test of actual scenarios.The main work of this paper is as follows:（1）Because the offloading task of the vehicle is sometimes too large,this paper proposes to divide the vehicle computing task into small dependent sub-tasks,which can be processed in parallel.Meanwhile,this paper proposes a delay and energy consumption model based on the task segmentation.A constrained multi-objective optimization model for computing task offloading in the Internet of vehicles was constructed,and a non-dominated sequencing genetic strategy was proposed to optimize the objective function.New non-dominated relations and constraints were proposed for computing task offloading in the Internet of vehicles.In addition,a series of experiments are carried out and compared with other offloading methods to prove the effectiveness of the proposed algorithm.Experimental results show that the proposed algorithm has better performance.（2）Due to the mobility of vehicles,the services they request should often be migrated between different MEC servers to ensure their stringent quality of service requirements.However,due to the instability of the movement,frequent migration will increase the cost and delay,so it is very challenging to design a good migration method.This paper studies the service migration of vehicles in the dynamic environment in the Internet of vehicles,and minimizes the completion time of service migration under the condition of meeting the cost of migration.We use deep reinforcement learning to construct an improved depth deterministic strategy gradient algorithm to optimize the cost and delay of vehicle task migration in the Internet of vehicles.At the same time,we use the centralized training distributed execution method to solve the high dimensional problem of vehicle task migration.A large number of experiments show that the proposed algorithm has the lowest migration cost and greatly reduces the service delay compared with other relevant algorithms.