Research On Seamless Handover For Time-Sensitive Autonomous Driving Services

Autonomous driving has become a highly valued field in recent years,and our country strongly supports and promotes its development.5G has become one of the key technologies in the field of autonomous driving due to its ultra-low latency characteristics,and time-sensitive networking(TSN)has the core technology to achieve highly reliable and deterministic data transmission,and has also become a key technology in the field of autonomous driving.The integration of 5G and TSN has further promoted the development of the field of autonomous driving.However,facing the uncertainty of the wireless side,the existing cell handover schemes are unable to meet the requirements of seamless handover of autonomous vehicles.On the other hand,in the face of large-scale time-sensitive autonomous driving services,the wired side of the 5G-TSN network convergence architecture is required to make reasonable resource scheduling to meet time-sensitive autonomous driving services and prevent packet loss due to insufficient link resources for switching demand.This paper focuses on the time-sensitive autonomous driving services,and conducts research on the seamless handover technology of the timesensitive autonomous driving services on the wireless side and the wired side,in order to improve the safety of autonomous driving.First,this paper deeply studies the methods for seamless handover of autonomous vehicles on the wireless side.Aiming at the time-sensitive autonomous driving services,analyzing their requirements on delay and reliability,a cell handover scheme based on coordinated multi point joint transmission(CoMP-JT)is proposed.At the same time,a modified proximal policy optimization(MPPO)deep reinforcement learning algorithm is proposed to comprehensively consider the performance of the base stations,and make a reasonable multi base stations connection strategy for autonomous vehicles.It is verified by simulation that the seamless handover mechanism based on CoMP-JT technology and MPPO algorithm proposed in this paper can realize zero mobility interruption times of autonomous vehicles during cell handover,and greatly reduce the intercell interference of autonomous vehicles at the cell edge,improving its average quality of experience(QoE)and average throughput.Secondly,this paper studies the resource pre-scheduling method of large-scale time-sensitive autonomous driving services on the wired side.Analyzing the applicable scenarios and existing problems of main timesensitive resource scheduling mechanisms of TSN,a services resource prescheduling scheme based on multi cyclic queuing and forwarding(MultiCQF)is proposed.The scheme is modeled and its constraints are established.To solve the resource scheduling problem,an algorithm based on simulated annealing(SA)algorithm is proposed.It is improved to add the memory function.It is verified by simulation that the resource prescheduling method proposed in this paper can deal with the scheduling problem of large-scale time-sensitive autonomous driving services traffic,and achieve a better solution within a limited time.In addition,in the face of time-sensitive autonomous driving services with strict data transmission requirements,the proposed resource pre-scheduling method can achieve a smaller average end-to-end delay.