Research On Power And Storage Management Of Vehicular Networks Based On Age-of-Information

The vehicular networks are attracting more and more researchers’ attention because they make traffic safer,more efficient and more intelligent.In order to reduce the incidence of traffic accidents and improve passenger experience,it is very important to ensure the freshness of information obtained by vehicles.This paper considers that vehicles acquire information from roadside units(RSUs).Then,the communication power and information storage allocation of RSUs are optimized to make the information obtained by vehicles fresher.At present,there are few researches on the communication power and information storage allocation based on multiple RSUs.Furthermore,most of them use traditional optimization algorithms.These algorithms will lead to high computational complexity when applied to complex vehicular network systems,and it is difficult to adapt to the dynamics of vehicular networks.In this paper,age of information(Ao I)is used to quantify information freshness,and a multi-agent reinforcement learning based method is proposed to minimize Ao I,including the following researches.A scenario of vehicles acquire information from the RSUs connected with storage servers is established,and the information is divided into different types according to time sensitivity and functions respectively.The channel model of communication between the RSUs and the vehicles and the queuing model of information transmission from RSUs to vehicles are constructed,and the expression of the total Ao I of the system is derived.Considering the scenario of time-sensitive information and time-insensitive information coexisting,this paper researches the Ao I based communication power control with reinforcement learning in vehicular networks.Subjecting to the minimum download rate requirement of vehicles and the maximum total communication power limitation of RSUs,a multi-agent twin delay deep deterministic policy gradient(MATD3)based method is proposed to minimize the total Ao I of the system by optimizing the communication power allocation of RSUs.The simulation results show that the proposed method has better convergency and reduces the total Ao I of the system than the methods based on deep deterministic policy gradient(DDPG)and greedy algorithm.Aiming at the scenario of paired RSUs,the communication power and information storage allocation method of vehicular networks based on Ao I and digital twin was studied.The digital twin system obtained the environment data in real time and constructed digital images,and controlled the fog servers to realize resource allocation of paired RSUs.Then an optimization problem is designed to optimize the RSUs communication power and information storage allocation to minimize the total Ao I of the system.Then,a MATD3 based method is proposed to solve the optimization problem.The simulation results show that the proposed method improves the Ao I performance of the system.In addition,the increasing of the number of vehicles,channels and information types will increase the total Ao I of the system.The above research proposes a communication power and information storage allocation scheme based on multi-agent reinforcement learning in vehicular networks,which effectively improves the freshness of information obtained by vehicles and provides a solution for the resource allocation problem in vehicular networks.