Intelligent Transmission Mode Selection And Computation Offloading In 5G Cellular Internet Of Vehicle

As an important part of future intelligent transportation system,the Internet Of Vehicle aims at providing stable communication links and safe data transmission for vehicles’ interconnection.With the rapid growth of number of vehicles and the progress of communication technology,in-vehicle applications have been developed quickly,and new applications such as automatic driving have emerged.The development of Internet Of Vehicle has requirements for data transmission and task processing,while the existing transmission mode and data processing ability cannot guarantee the quality of service for IoV.Firstly,the poor quality of data transmission is mainly caused by the single transmission mode of vehicles.While in the real vehicular network,there are a variety of transmission modes for vehicles,such as vehicle to vehicle(V2V),vehicle to infrastructure(V2I)and vehicle to network(V2N).V2V mode can provide low latency data transmission,but its bandwidth is too small to gauarantee large data transmission.In contrast,the bandwidth of V2I mode is larger,while its coverage is limited and the cost is higher.V2N has larger coverage and bandwidth,but its quality of service is limited by distance.Thus,the use of a single transmission mode will have a number of defects and is unable to meet the diverse needs of vehicles.Secondly,the amount of data collected by in-vehicle applications is huge,and the local computing resources of vehicles are limited,which is not enough to support large computation.As a supplement of computing resources,a single edge computing node can only provide limited computing resources.The cloud edge collaborative computing node solves the problem of computing resource shortage,but increases the transmission delay of the system.In order to sovle the problems of data acquisition and processing,this paper proposes an intelligent transmission mode selection method for 5g cellular Internet of vehicles to meet the diversified needs of data transmission,and proposes a computing task offloading method based on edge server cooperation to solve the problem of insufficient vehicle computing resources.1)Intelligent transmission mode selection method of 5G cellular Internet of vehicles based on online reinforcement learning.First,based on the characteristics of different transmission modes in cellular vehicle network,the vehicle can choose three transmission modes to ensure high reliability and high performance data transmission links.Secondly,the auto-encoder based on graph neural network is used to cluster the vehicles,and the vehicles in the same cluster can share the information.Secondly,a transmission mode selection algorithm based on Multi-Agent Reinforcement Learning is proposed.The agent dynamically adjusts the strategy by interacting with the environment,which effectively improves the performance of mode selection in large-scale Internet of vehicles.Simulation results show that,compared with spectral clustering and K-means clustering algorithm,graph auto encoder has better clustering effect.Compared with random algorithm and no intelligent selection algorithm,the proposed algorithm has higher system throughput.2)A 5G cellular Internet of vehicles task offloading method based on multi distributed MEC collaboration.Firstly,the cooperative characteristics of 5G cellular network are used to build a multi distributed MEC cooperative 5G cellular vehicle network.The vehicles in the network will perform computing intensive tasks.Secondly,the computation offloading problem is decomposed into spectrum resource allocation problem and task processing decision problem.Due to the limited spectrum resources,different cells share spectrum resources.In order to reduce the interference caused by the same frequency scheduling,this paper uses bipartite graph matching algorithm to determine the channel allocation scheme.Based on the requirements of energy consumption and delay in the Internet of vehicles,an online Lyapunov optimization method is proposed to solve the problem of computing resource allocation and achieve the dynamic balance of time delay and energy consumption.Finally,simulation results show that the proposed multi distributed MEC cooperative architecture has better performance in terms of energy consumption and execution delay.