Research On Key Technologies Of Mission Offloading For Vehicular Edge Computing

The Internet of Vehicles realizes the information exchange between vehicles and the outside world with the help of a new generation of information and communication technologies.The popularity of intelligent transportation has given new life to the Internet of vehicles.Various new mobile applications with low delay requirements,such as autonomous driving and adaptive navigation,have swarms into the Internet of vehicles.Mobile edge computing is regarded as a powerful example to deal with the growing demand for real-time computing and the limited computing resources of users.As an innovative technology combining the Internet of vehicles and mobile edge computing,vehicular edge computing can meet the needs of low delay,low energy consumption and high reliability of on-board communication.As one of the key technologies of mobile edge computing,computing offloading can migrate the computing process from mobile devices to servers with rich resources.However,the continuous dynamic changes of traffic environment and the intensive task data in the Internet of vehicles make computing resources limited and network load unbalanced.How to divide on-board tasks according to environmental factors and issue targeted offloading decisions is the key to efficient utilization of system computing resources.At the same time,vehicular edge computing needs to be oriented towards the unsteady scenario to solve the task continuity problem caused by vehicle migration and zone crossing.In addition,to evaluate the pros and cons of the uninstall scheme,it is necessary to consider not only the overhead of the client side,but also the energy consumption of the server side.Therefore,this paper carries out the following research on the edge offloading technology oriented to the Internet of vehicles:1.Deep learning-based task discrimination offloading in vehicular edge computing.This scheme can classify the on-board task types in advance and provide specific offloading decisions.To be specific,firstly,dynamic vehicle characteristic parameters were selected as criteria,and the on-board tasks were divided into three categories: time-delay sensitive,energy consumption sensitive and non-sensitive by using improved analytic hierarchy process(AHP).Then the joint modeling of resource allocation is carried out based on three kinds of offloading decisions and the constraint of modeling is eliminated by scheduling algorithm and penalty function.Finally,a distributed offloading network based on deep learning is proposed to effectively reduce energy consumption and delay of vehicle edge computing system.The experimental results show that the proposed offloading scheme has good adaptability to traffic environment and effectively reduces the time delay and energy consumption of task processing.2.Quality of Service optimization offloading in mobility scenario based on vehicle edge computing.The scheme can ensure the continuity and reliability of the service and provide a server deployment scheme.Specifically,the center edge server is built based on the overcrossing movement scene of vehicles to synchronize the information of each edge server and deliver the unfinished tasks.Then an improved particle swarm optimization algorithm is proposed to obtain the optimal position of task offloading.Finally,the deep neural network is used to process the output results of the previous step to obtain the optimal offloading strategy and resource allocation scheme.Experimental results show that the proposed offloading scheme has better service quality performance and minimizes the cost of vehicle edge computing system.3.Service pricing computing offloading based on non-cooperative game in vehicular edge computing.In this scheme,the server side cost is taken into account and service pricing is used to solve the distributed task offloading problem.Specifically,in the multi-vehicle user and multi-server scenario,the pricing modeling is carried out for the cost expenditure of the client side and the service income of the server side in the process of task offloading.Then,the overall revenue objective function of the Internet of Vehicles system was transformed into a potential game equation,and the optimal offloading strategy was approximated by iterative updating of game strategy.Finally,the algorithm in this chapter is simulated.Experimental results show that the proposed potential game equation can converge to Nash equilibrium and can effectively improve the overall benefits of VEC system.