Research On Dynamic Resource Allocation Algorithms In Vehicle-to-everything Networks

In recent years,transportation issues from travel,safety and environment have made Vehicle-to-Everything Networks a key topic of research at home and abroad.As a potential Vehicle-to-Everything Networks,Cellular Vehicle-to-Everything(C-V2X)has become a research hotspot.Vehicle-to-vehicle(V2V)communication,as one of the important communication types of C-V2X,is a key technology to improve traffic efficiency and reduce traffic accidents.This paper studies the resource allocation of V2V communication based on Device-to-Device(D2D)technology when multiplexing the uplink of cellular users.Considering the frequent changes of the channel state caused by the high-speed mobility of the vehicle,a dynamic resource allocation algorithm of V2V is proposed.Aiming at the dynamic resource allocation problem at millisecond level,a V2V dynamic resource allocation algorithm based on multi-agent deep Q-learning is proposed.Under the condition of ensuring the reliability of V2V,each V2V can allocate resources distributed with the movement of vehicles,so that the total throughput of cellular users in the system is maximized.In this algorithm,each V2V is an agent and learns its DQN through continuous interaction with the environment to maximize the total throughput of cellular users.Simulation results show that the algorithm is effective,that is,the resource allocation of V2V through its convergent DQN can guarantee the reliability of V2V communication and the communication performance of cellular users.Aiming at the dynamic resource allocation problem at the level of 100 millisecond,this paper proposes a V2V pre-resource allocation algorithm based on large-scale fading.The base station can calculate the large-scale fading in the future time through the predictability of vehicle trajectory,and use this algorithm to allocate resources for V2V in advance,so as to realize the dynamic resource allocation of V2V.The algorithm is divided into four steps to execute.First,calculate the channel power gain in the cellular vehicle networking in the future according to the predictability of the vehicle trajectory;Then,use the interference channel power gain to cluster the V2V user pairs with less interference,and all V2V of each cluster communicate in the same spectrum;Then,pair the clustered user groups with cellular users one-on-one,and use the proposed power allocation algorithm based on Q learning to find the V2V power allocation method that maximizes the throughput of cellular users during pairing;Finally,use the matching algorithm in graph theory to find the best match for user groups and cellular users.In the simulation,the algorithm is compared with the static resource allocation algorithm and the random resource allocation algorithm,which proves that the algorithm has a certain degree of effectiveness and can guarantee the reliability of V2V communication and the communication performance of cellular users.