Research On Task Offloading Strategy Based On Mobile Edge Computing In Internet Of Vehicles

With the development of science and technology,as well as the support of policy,the number of intelligent connected vehicles(ICV)is gradually increasing.With the rapid development of vehicle applications,the computing power of vehicle computing units is gradually inadequate.It is considered to be a promising method for ICV to offload tasks to the cloud server through V2 X link.However,the long-distance deployment of cloud server and unstable backhaul link bring high delay fluctuations,resulting in poor service experience of ICVs.As a new computing paradigm,mobile edge computing(MEC)sinks computing resources to the edge of the network to improve the quality of service.In this thesis,MEC is used to meet the requirements of time sensitive task of ICVs.The research idea is to use the available edge nodes near the vehicle to enhance the computing power of the task vehicle.The research contents and contributions of this thesis are as follows:Firstly,the task offloading algorithm based on Stackelberg game is studied.The main optimization objectives are task delay and cost.In this thesis,a local-edge collaborative computing model is constructed,which considers user psychology,vehicle speed,acceleration,location,communication ability and computing ability.According to the task delay,expenditure and user sensitivity,the utility function of task vehicle and service vehicle is established.The service vehicle maximizes the revenue by pricing the computing resources,and the task vehicle adjusts the strategy to meet the user experience.This thesis uses the Stackelberg game model to model the interaction between the task vehicle and the service vehicle,and proposes a task offloading algorithm based on the Stackelberg Loop-Iteration algorithm in the Internet of vehicles(Io V)environment.It proves that there is a Nash equilibrium between the service vehicle and the task vehicle.Experiments show that the algorithm achieves a balance between task vehicle delay and expenditure,task vehicle utility and service vehicle utility under various constraints,and has higher performance than other algorithms.Then,the task offloading algorithm based on reinforcement learning is studied.This thesis combines the advantages of cloud computing model and mobile edge computing model,and introduces cloud server as backup to alleviate the shortage of computing resources when the computing power of edge nodes is insufficient.On the basis of previous research,this thesis increases the impact of task priority and task success or failure on the utility value,and constructs a local-edge-cloud offloading model of three-layer Stackelberg game,Stackelberg-MADDPG algorithm is proposed to solve the problem.The simulation results show that the algorithm proposed in this thesis achieves a higher total system utility,improves the task success rate under various constraints,and achieves a balance between task vehicle delay,expenditure and task success rate,task vehicle utility and service vehicle utility,and has better performance than other algorithms,and the execution time is extremely short.This thesis has 28 figures,4 tables and 89 references.