Computation Offloading Scheduling In Vehicular Networks Based On Mobile Edge Computing

As an essential aspect of human survival and development,vehicles and road infrastructure are stepping into a new era of information and intelligence with the coming of 5G.The rapidly developing vehicular networks will provide communication hardware support for autopilot technology and intelligent transportation system(ITS).Simultaneously,vehicular applications and services become more and more diverse day by day,hence,the users' requirements on data rates and quality of service are exponentially increasing,making the limited hardware resources be the bottleneck.The shortage of computing and storage resources on vehicles will be solved and the requirements of ultra-reliable low latency connection(uRLLC)will be satisfied if mobile edge computing(MEC)is introduced into vehicular networks.How to schedule the offloading of computation tasks in vehicular networks with MEC is a subject worth studying.Firstly,the paper formulates a unidirectional and one-dimensional model of road considering the deployment of roadside units(RSUs)and MEC servers and the distribution of vehicles.On the one hand,based on time consumption,this paper considers a case when the speed of vehicle is too high or the computing task is too heavy,which causes that the vehicle has been away from the coverage of the RSU when the task offloaded to MEC severs is finished.Generally speaking,the downloading of computing output will be transmitted between RSUs through wireless backhauls which may jitter.Besides,the waiting time of a task may be long while the load of MEC server is rather high.The paper proposes a efficient predictive offloading scheduling strategy with vehicle-to-vehicle(V2V)communication based on MEC servers' load state,which save not only the transmit time through wireless backhaul by uploading and downloading data via V2 V communication but also the waiting time at MEC server by selecting the least load MEC server to offload task.On the other hand,energy consumption is also a key factor that should be considered when offloading computation task.The paper defines the weights of completion time and energy consumption from a vehicle's perspective,then the costs when task is executed on local vehicular terminal and MEC server are expressed.For the two ways,the paper calculates best CPU computing frequency and best antenna transmission power by convex optimization respectively.Then,the lowest costs on local terminal and on MEC server are calculated,by which the vehicle can make the best choice.