SDN-based UAV-assisted Cost Optimization Strategies For Vehicular Computing

With the rapid development of vehicular network,people’s demand for vehicular network application is also growing rapidly.However,most of these applications are delay-sensitive and compute-intensive.At present,there are still large numbers of legacy vehicles in many areas.The storage capacity,computing resources and battery life of these legacy vehicles are limited.Also,some areas are underdeveloped,with less infrastructure and intensive construction.These factors limit the legacy vehicles to perform these applications on the road.In recent years,edge computing has become a popular computing offloading intelligent device with its powerful computing power and convenience to provide users.However,due to the obstruction of dense buildings or the lack of infrastructure in some areas,it is difficult to guarantee the communication quality of computation offloading.With its small size and easy deployment,UAV has become one of the means to establish communication links at both ends.Therefore,in order to make a reasonable computation offloading decision,we design an optimization scheme of UAV assisted vehicle computing offloading based on software defined network architecture to minimize the system cost of vehicle computing tasks.In this scheme,the UAV and the mobile edge computing server can execute the computation tasks on behalf of the vehicle users.At the same time,the UAV can also be deployed as a relay node to help forward computation tasks to the MEC server.We formulate the offloading decision-making problem as a multi-user computation offloading sequence game,and design a UAV-assisted vehicular computation cost optimization（UVCO）algorithm to solve this problem.Furthermore,we also consider offloading part of the vehicular computation tasks to other intelligent devices that can establish communication connections,so that multiple devices and vehicle can execute computation tasks in parallel,making full use of the computing resources in the system,and greatly improve the execution efficiency.In order to better allocate computation tasks to multiple devices,we propose a minimum incremental task allocation（MITA）algorithm.Simulation results show that the SDN based computation offloading decision optimized by our two algorithms UVCO and MITA greatly improve the efficiency of legacy vehicle computation tasks,and greatly reduce the cost of executing computation tasks,making it possible for legacy vehicles to execute delay-sensitive and compute-intensive tasks.