Research On Network Resource Scheduling In Vehicle Infrastructure Cooperative System

The Internet of Vehicles(IoV)is a controllable,manageable,operable,and credible open converged network system based on the coordination of people,vehicles,and the environment.With the help of the new generation of commu-nication and information processing technology,the vehicles and people,roads,and services are connected through ubiquitous networking technology,enhancing the sense,cognition and decision level for autonomous driving,and ultimately achieving deep integration between people,vehicles,and the envi-ronment.Intelligent vehicle infrastructure cooperative system(CVIS)is an im-portant scenario for IoV.Through the comprehensive implementation of vehi-cle-vehicle and vehicle-road dynamic real-time information interaction,vehicle active safety control and road collaborative management are carried out to fully realize the effective coordination of people and vehicles,ensure traffic safety,and improve traffic efficiency.Due to the characteristics of vehicle mobility and uneven distribution in time and space,IoV will face resource bottlenecks in supporting large-scale vehicle-road collaboration.This paper is concentrated on the resource bottleneck problem in the Cellular-V2X network in guarantee-ing the delay,reliability,bandwidth and other requirements of IoV service,fo-cusing on scheduling the IoV network resources from media access control layer,network layer,and application layer.This dissertation achieves the fol-lowing contributions and main results:1)At the media access control(MAC)layer,in terms of the contradiction between stable resource allocation and vehicle high mobility,a software-de-fined network(SDN)inspired MAC protocol is proposed to support the safe driving-related service,which requires low latency,high reliability,high-fre-quency cooperative communication between vehicles and roads within the lo-cal area.Based on the SDN technology,a two-tier architecture is first proposed to support vehicle-vehicle and vehicle-road collaboration;then,the roadside infrastructure centrally allocates and manages slot resources based on shared slot allocation messages to ensure reliability broadcast;further,in order to solve the problem of vehicle density changes,an adaptive communication range broadcast mechanism is designed.Finally,a theoretical model is proposed to analyze the performance of the distributed MAC protocol under imperfect channels.The joint simulation results based on traffic simulation software and communication simulation software show that the proposed method achieves obvious performance improvements in terms of access delay,access collision rate,and network throughput.2)At the network layer,in terms of the efficiency of data cooperative trans-mission,a network topology reconstruction method based on graph theory is proposed to support the cooperative sharing of large bandwidth and low latency data.The proposed method first adopts the cellular link to collect vehicle-road context information,and allows the base station to schedule device-to-device direct communication between vehicles.Then,the cooperative data transmis-sion problem is modeled as the maximum independent subset problem based on the conflict graph.In the constructed conflict graph,the vertices on the graph represent possible transmissions,and the edges mean wireless communication restrictions between transmissions.Finally,an equilibrium greedy algorithm and an integer decomposition algorithm are proposed to solve the maximum independent subset problem for the efficient reconstruction of the network to-pology.The joint experimental results based on traffic simulation software and communication simulation software show that the proposed scheme can im-prove the efficiency of collaborative data transmission while reducing com-plexity.3)At the application layer,a large-scale vehicle-cellular communication load balancing method based on global traffic situation learning is proposed to support the balanced distribution of driving data.First,mesh the considered area,and apply the inflow and outflow of the grid to represent the traffic situa-tion.Secondly,a three-dimensional convolutional neural network is designed to effectively learn and mine the temporal and spatial correlations contained in the traffic situation,and based on this,the evolution trend of the traffic situation is predicted.Finally,the load balancing problem is modeled as a nonlinear op-timization problem,and a convolutional neural network is used to approximate the optimization process.Experimental results based on the three-month taxi and bus trajectory data in Beijing show that the proposed scheme can improve the utilization rate of network resources.