Research On Optimization Method Of VANET Based On Graph Embedding

Vehicle ad hoc network(VANET)is widely recognized as a technology to improve road traffic efficiency and driving safety.It uses each vehicle as a source of information,and uses wireless communication means to establish an information system of human,vehicle and road with vehicles as nodes.However,compared with traditional cellular networks and mobile private networks,VANET is influenced by human factors,has high network dynamics and complexity,as well as poor channel conditions,making it difficult to build a reasonable and stable vehicle network.In response to these issues,this dissertation first investigates effective methods for extracting vehicle features,then constructs a reasonable dynamic network topology using vehicle features,and finally studies methods for improving communication conditions and core nodes deployment strategies of the network.This achieves low latency and cost vehicle networking with good robustness,connectivity,and dynamism.The research results mainly include:(1)Vehicle feature extraction method based on fuzzy reasoning.Collect and process vehicle information,then use fuzzy reasoning methods,combined with actual vehicle driving logic,to determine the relative driving style of the driver,and map the vehicle into a low dimensional feature vector.(2)Topology organization of vehicle network based on graph embedding.Propose the label-range graph embedding method based on the characteristics of VANET,use the cold start method to modify the feature vectors of newly added network nodes,random walk between vehicle nodes,and optimize the entire network topology.Finally,obtain a network topology with good dynamics,connectivity,and robustness.(3)RIS based software definition network deployment method.Set specific vehicle nodes as virtual central nodes for distributed control.Optimize the deployment of reconfigurable intelligence surface,and propose a dynamic deployment mechanism of central nodes,which can optimize the channel capacity by adaptive dynamic deployment according to the channel state.