Research On Task Offloading Strategy Based On Vehicle Edge Computing

With the development of 5G technology,the "Internet of Everything" has been further promoted,and new computing-intensive applications such as autonomous driving have emerged in due course.These new applications occupy a large amount of computing and storage resources when processing tasks.Real-time processing poses a huge challenge.Traditional task computation offloading modes,such as cloud computing and fog computing,are difficult to efficiently perform computing-intensive tasks,resulting in a decline in user experience quality.To alleviate this difficulty,Mobile Edge Computing(MEC)came into being.MEC can offload computing tasks to network edge nodes,reduce data processing delay and reduce terminal energy consumption.Introducing MEC into the Internet of Vehicles can meet the expansion requirements of vehicle computing capabilities,but also make up for the long delay of cloud computing and energy consumption.Due to the disadvantage of high power consumption,MEC is considered to be one of the key technologies to solve vehicle communication.However,MEC server resources are limited and may not be able to meet the offloading requirements of computing tasks.Therefore,it is very important to design a reasonable offloading strategy and resource allocation scheme considering the characteristics of each computing platform and task.Based on existing research,this thesis conducts research on task offloading and resource allocation of Vehicular Edge Computing(VEC).The main work is as follows:1)The overall task offloading strategy in the vehicle edge computing scenario is studied.The design goal is to minimize the system cost under the constraints of the maximum tolerable delay,computing resources and task types.Since computing tasks are inseparable,relying solely on edge devices is insufficient in computing power,so it is necessary to use limited computing resources to reasonably allocate task offloading strategies.In order to minimize the system cost,the optimization problem is modeled as a mixed integer nonlinear programming problem with constraints,and it is decomposed into two sub-problems of task offloading decision and resource allocation.Aiming at this problem,a task offloading algorithm based on discrete binary particle swarm(Binary Particle Optical Swarm,BPSO)is proposed.The BPSO algorithm is used to solve the 0-1 integer programming problem of the task unloading decision,and under the specific unloading decision,the standard particle swarm optimization(Particle Optical Swarm,PSO)is used to optimize the computing resources.In order to simplify the solution,the penalty function method is used to construct the particle fitness function.At the same time,based on the channel selection algorithm of idle vehicles,an effective task offloading strategy is proposed to achieve a reasonable allocation of offloading ratio and MEC offloading resources.The simulation results show that compared with other traditional benchmark algorithms,the offload algorithm proposed in this chapter can meet the maximum tolerable delay and minimize the total system cost.2)Considering that Non-Orthogonal Multiple Access(NOMA)technology can improve network spectrum efficiency,allow multiple vehicles to share the same wireless resources,and reduce network congestion,this thesis studies the vehicle edge computing network based on NOMA resource allocation and partial offloading of vehicle user tasks.Under the condition of satisfying the vehicle’s maximum delay tolerance threshold,power,computing resources and other constraints,the design goal is to minimize the system cost and realize the comprehensive coordination and resource allocation of network resources.Since the problem is non-convex,the solution is too complex.Therefore,a penalty function is constructed,and all optimization variables are regarded as a particle,and a multi-user task offloading strategy based on Adaptive Particle Swarm Optimization(APSO)algorithm is proposed.Since the constraints may be violated when particles are updated,the proposed method handles the particle out-of-bounds situation specially.This method can adaptively adjust the inertia weight to improve the optimization ability.The simulation results show that the performance of the proposed offloading method is better than that of local computing,standard particle swarm and other comparison schemes,and can effectively reduce the total cost of the system.