Research On Edge Computing Resource Scheduling Strategy For Internet Of Vehicles

With the rapid development of Io T and 5G technology,Telematics has become one of the landmark application scenarios of 5G technology,driving the generation of a large number of emerging in-vehicle intelligent applications.However,the limited computing resources of vehicle terminals make it difficult to meet the requirements of latency and energy consumption for emerging applications such as real-time road conditions and intelligent identification.The emergence of edge computing-based Telematics technology solves the above problems.Invehicle edge computing is to deploy edge servers to Road Side Units(RSUs)closer to vehicles to provide computation and storage services for vehicles,relieving the computation and storage pressure of vehicle terminals.However,due to the limited resources of in-vehicle edge computing,the unlimited scheduling of tasks to the edge servers can lead to overloading of servers and affect the quality of service(Qo S)level of vehicle users.A reasonable resource scheduling strategy can ensure the comprehensive performance of the edge network and improve the Qo S of users.therefore,this paper addresses these issues by doing the following research:(1)A two-stage binary resource scheduling strategy is proposed for the problem of limited computational resources in the in-vehicle edge computing environment and for non-splittable tasks.The binary resource scheduling problem is decomposed into two factors,offloading decision and resource allocation,with the objective of minimizing the linear combination of delay and energy consumption.The optimal offloading decision is derived by an improved fusion genetic algorithm in the first stage,and the optimal resource allocation is solved by an improved artificial fish swarm algorithm in the second stage,and the two are jointly optimized to reduce the system overhead.Simulation results show that the proposed strategy can effectively reduce delay,energy consumption and overhead compared with other benchmark strategies.(2)A partial resource scheduling strategy based on an improved simulated annealing algorithm is proposed for the problem of complex road traffic and dense vehicles generating too many computational tasks that need to be processed in real time in vehicular networks,and for detachable tasks.With the objective of minimizing the system overhead,a partial scheduling system model with an offloading ratio factor is constructed.The partial resource scheduling problem is decomposed into two factors: the unloading ratio and the computational resource allocation.The optimal unloading ratio factor is solved by an improved simulated annealing algorithm,and the optimal resource allocation is derived by the Lagrange multiplier method during the iteration of the algorithm.Simulations verify the convergence of the proposed strategy and its efficiency in optimizing delay,energy consumption and overhead.(3)Since intelligent applications for vehicles generate oversized computational tasks,a finegrained resource scheduling strategy that slices tasks into subtasks with dependencies is proposed for detachable tasks.Taking in-vehicle navigation applications as an example,the navigation tasks are split and processed in parallel,which can significantly reduce the system latency.With the goal of minimizing task completion time,a system model for fine-grained subtask splitting is constructed.Firstly,for single-vehicle single-server,this paper proposes the ESF(Earliest Scheduling Finish time,ESF)algorithm,which integrates the task priority and processor selection problems.Secondly,for multi-vehicle multi-server,the distributed DESF(Distributed Earliest Scheduling Finish time,DESF)algorithm based on game theory is proposed in this paper.The convergence of the proposed algorithm is verified by simulation and the system delay can be effectively optimized.