Priority-based Task Assignment Offload Algorithm For Vehicle Edge Network

With the expansion of information technology and new business requirements,a huge number of automotive applications with requirements for computational power and storage resources have developed as one of the typical application scenarios of 5G mobile communication.In today’s vehicle communication environment,however,traditional task migration and resource allocation approaches struggle to fulfill the low latency and high reliability needs of modern car applications like real-time road conditions and intelligent recognition.In new vehicle application contexts such as intelligent transportation,mobile edge computing(MEC)may deploy a large number of high-performance servers on the network edge side.Terminal cars dump tasks to the MEC edge server for execution,reducing the time it takes for various activities to complete.As a result,the focus of this research is on vehicle ad hoc network communication in an edge environment,and it presents two alternative task offloading methods based on deep reinforcement learning(DRL)for various application scenarios.(1)This paper proposes a delay optimized task unloading algorithm based on deep Q network(DQN)task priority classification for the one-way Lane multi terminal vehicle application scenario,which allows the terminal vehicle to maximize task processing rate while reducing the average delay of user tasks.To ensure the processing efficiency and service quality of emergency activities,the analytic hierarchy process(AHP)is utilized to split the priority of terminal vehicle services,with a larger weight coefficient given to the delay constraint and security level.Second,according to the task preprocessing time,initialize and remove the service.Finally,an acceptable computing task offloading strategy is constructed using the task priority design decision-making approach.According to simulation findings,refind the algorithm successfully boost the total income of the system,minimize the average latency of user actions,and improve the end user’s use experience.(2)This study offers a task offloading technique for edge computing of the Internet of cars based on vehicle clustering for the task scenario of vehicle cluster,taking into account that the increase in the number of vehicle terminals would pose certain obstacles to MEC server nodes.To implement the Internet of vehicles user clustering,first a system model for the Internet of vehicles cluster scenario is built,and then the K-means clustering method with user priority as the clustering feature is employed.Edge computing and local computing are assigned to the clusters with the greatest and lowest user priority,respectively,through user clustering.DQN is used to learn the best distributed computing offloading mechanism for average cluster users.The suggested approach significantly decreases the overall execution latency of system tasks,according to simulation findings.