Research On Low-latency Edge Computing Task Offloading And Relay Algorithm For Internet Of Vehicles

In recent years,the Internet of things(IOT)has been a hotspot in academic and industrial research.With the rapid development of the IOT,more and more devices can be connected to the wireless communication network,which makes the communication between devices more complicated.Vehicles can form a Vehicle Edge Network(VEN)by communicating with other vehicles or roadside infrastructures.As an important part of intelligent transportation,VEN not only provides real-time information for vehicle driving but also provides convenience and security for our life.However,due to the dynamic topology change of vehicle nodes and the coexistence of heterogeneous nodes,it is difficult to provide stable and timely information transmission service for all communications.Therefore,in order to solve the problem mentioned above,this paper carried out research on task offloading and data relay in VEN.The main work and innovations are as follows:(1)Due to the high-speed mobility of vehicle nodes in VEN,the network topology changes frequently,resulting in the frequent failure of data transmissions.This paper proposed a relay vehicle selection scheme based on D-S evidence theory and Long Short-term Memory.Firstly,the untrusted vehicle nodes are eliminated based on historical credibility and residual energy.Then,the vehicles that meet current credibility requirements are selected as candidate relay vehicles according to the D-S evidence theory.In addition,each candidate vehicle predicts its own driving trajectory based on historical data using Long Short-term Memory(LSTM)network.Finally,the proposed relay vehicle selection algorithm is adopted to select the relay vehicle transmission data.Experimental results show that this method can effectively predict the vehicle trajectory and select the vehicle consuming the shortest time,which ensures the stability of the transmission link and reduces the transmission delay.(2)In order to solve the problem of latency increase caused by task offloading and unreasonable resource allocation,this paper proposed a task offloading and resource allocation scheme based on an evolutionary algorithm.Firstly,the latency and energy consumption of the network are completed from two perspectives of vehicles and the MEC edge server.Then,a new cost function is defined to represent the cost of completing the task by considering the proportion of the latency and energy consumption comprehensively.Moreover,the solution of the cost function is transformed into a mixed integer nonlinear programming problem(MINLP).Finally,the evolutionary algorithm is utilized to optimize the MINLP and find the optimal solution by initializing the population,mutation,crossover,calculating individual fitness and selection.Experimental results show that the optimal proportion of task offloading and the most reasonable scheme of resource allocation can be obtained based on the real-time task and resource situation.The transmission delay and calculation delay of mission can be reduced based on the proposed methods,which improve the quality of service.