Research On Perception Scheduling And Task Offloading Strategy Based On Cooperative Vehicle Infrastructure System

With the development of the Internet technology,the emergence of the Internet of vehicles has prompted the automobile to gradually enter the era of intelligent driving,and Cooperative Vehicle Infrastructure System(CVIS)connecting cars and the infrastructure is the technical research focuses in the field of the future Internet of vehicles.By sharing the perception information and computing resources at both ends of the vehicle and the infrastructure,CVIS can help vehicles to obtain road environmental information and provide computing services for them,effectively improving the perception ability of vehicles and reducing the computing delay of perception tasks.Therefore,based on CVIS,this thesis studies the perception scheduling and task offloading in this scenario.In existing studies,CVIS usually sends environmental data from all roadside units to vehicles.However,the excessive environmental information cannot improve the perception ability of vehicles correspondingly,and will cause the waste of system resources and increase the system delay,which is difficult to meet the real-time requirement of CVIS.To solve this problem,on the basis of ensuring the perception needs of vehicles,this thesis proposes a roadside units’ perception scheduling and resource optimization allocation strategy based on CVIS,which can reduce system delay and resources waste by scheduling some roadside units to transmit perception information to vehicles.Specifically,a joint optimization model with the goal of minimizing the system delay is established by first considering scheduling decision and resource allocation of the roadside units and the perception demand of vehicles.Then,the optimization model is modeled as a Markov decision process,and we propose a perception scheduling and resources optimal allocation of roadside units algorithm based on DDPG to solve it.The simulation results show that the proposed algorithm can effectively improve the efficiency of resources allocation and reduce the system delay under the condition of meeting the perception requirements of vehicles.Further,aiming at the problem of environment perception and task offloading of intelligent vehicles in CVIS,this thesis proposes an offloading scheduling strategy for vehicle-infrastructure collaborative perception tasks.By controlling the offloading and scheduling of perception tasks of vehicles and the infrastructure,the balance among average delay,average energy consumption,average perception accuracy and average perception range of vehicles in CVIS is achieved.Specifically,considering the perception scheduling decision of vehicles and the infrastructure,offloading decision and transmission power allocation of vehicles,an optimization model is established to balance the average tasks processing delay,average energy consumption,perception accuracy and perception range of vehicles in CVIS.Then,the optimization model is modeled as a Markov decision process,and an offloading scheduling and resource allocation algorithm based on DDPG is proposed to solve the optimization problem.The simulation results show that under different weight parameters,the proposed algorithm can effectively realize the tradeoff among the average delay,average energy consumption,average perception accuracy and average perception range of vehicles.Moreover,compared with other contrastive algorithms,the proposed algorithm has better performance,which verifies the effectiveness of the proposed offloading scheduling strategy for vehicleinfrastructure collaborative perception tasks.