Research And Implementation Of Edge Computing Resource Optimization In Vehicle-Road Collaborative Environment

With the continuous development of 5G technology and intelligent technology,intelligent networked vehicles can rely on vehicle-road collaboration technology to form a more specific understanding of the road traffic situation through the edge cache of the roadside infrastructure and edge computing services to ensure driving safety and improve traffic efficiency.Due to the complex conditions of the transportation system and the diverse vehicle requirements,the communication,cache,and computing resources provided by the roadside infrastructure for intelligent networked vehicles usually result in unbalanced spatial and temporal distribution of demand and unbalanced resource supply and demand,which affects service quality.There is a high degree of coupling between the route distribution of vehicles and the regional requests of edge resources in the transportation system.The effect of only balancing the load of edge resources through service placement and migration is limited,and traditional centralized scheduling strategies usually encounter computational bottlenecks due to the large scale of traffic system problems.In order to cope with the above problems,we use multi-agent deep reinforcement learning to solve the problem of joint optimization of "Road and Edge service" resources in the scenario of vehicle-road collaboration.In this paper,the road resources,communication resources,network resources,and computing resources in the vehicle driving and service migration scenarios in the vehicle-road collaborative environment are modeled and optimized goals are defined,then based on the MADRL method,a dual-channel network model is proposed to help intelligent networked vehicles make joint optimization decision of driving planning and service migration to realize joint optimization scheduling of road resources and edge computing resources.We design simulation experiments to prove that the algorithm in this paper can use the road resources and edge resources in a balanced way to ensure that the vehicle application meets the service delay requirements and ensure the efficiency of the drving.In the vehicle-road collaboration scenario,the movement mode of the vehicles not only affects the sequence of connected base stations,but also affects the communication quality of the connected base stations.These factors affect the scheduling decision of edge cache resources.Moreover,the cache scheduling method based on state snapshots is difficult to ensure the optimization of long-term utility.In order to solve this problem,this paper studies the problem of using small base stations to provide caching services for vehicles in a vehicle-road collaboration scenario.Finally,based on the long-term utility optimization of edge cache resources,a cache content distribution decision algorithm is proposed by using the idea of deep reinforcement learning.The algorithm outputs decision actions by evaluating the system state value and different action dominance function value.Simulation experiments are designed to prove that the algorithm can ensure the vehicle to obtain the target content in time and improve the utilization efficiency of cache resources.