Research On Resource Allocation Of Millimeter Wave IoV Based On Online Learning Of Multi-armed Bandit

With the increasingly extensive application of Internet of Things in the automotive industry,the characteristics of future Internet of Vehicle(IoV)communication links with high bandwidth,high transmission reliability and low delay also put forward higher requirements on the existing spectrum resources and network capacity.Compared with the crowded traditional low-frequency band,the millimeter wave(mm Wave)(30～300GHz)band with rich spectrum resources can provide higher transmission bandwidth capacity and communication rate,but its highfrequency characteristics will lead to high path loss and penetration loss.In addition,the dynamic nature of vehicular scenarios raises the complexity of directional mm Wave vehicular communications,and the inaccurate beam selection will also cause blocking loss and lead to the performance degradation of mm Wave communication system.Therefore,mm Wave vehicle-mounted communication system should have high environmental adaptability and context awareness ability so as to timely select accurate beam for mm Wave communication link.Existing mm Wave beam selection schemes based on Multi-armed Bandit(MAB)online learning are difficult to select a more reasonable beam set with width and direction due to the deficiency of context information and beam setting,which hinders the further improvement of the vehicular network capacity.There is also a lack of consideration of the energy efficiency of the Internet of vehicles network,such as base station energy consumption and power adjustment.The main work of this paper is as follows:(1)Aiming at the problem that mm Wave communication is prone to be affected by inaccurate beam selection in highly dynamic vehicle scenarios and performance degradation due to the easy blocking characteristics of mm Wave communication,a network capacity-oriented resource allocation scheme for mm Wave vehicular network based on online learning MAB is proposed.By setting more beam widths and beam directions,the mm Wave base station adapts to the dynamic communication environment and selects a more reasonable beam set by automatically detecting and learning the relationship between available vehicle context information and previous beam selection results.In addition to allowing beam overlap,the proposed scheme also uses more dimensional vehicle context information to help identify the desired better beam.In order to reduce the cost of beam exploration,the coverage area of the base station is divided into several sub-coverage sectors,and then the optimal beam set is selected from the selection results of sub-sector beam set.For beam performance update,weight robust update can better adapt to the change of communication environment.Simulation results show that this scheme increases the detection cost slightly and converges to the optimal solution,which greatly improves the network throughput of the system.(2)Aiming at the energy efficiency and power consumption of mm Wave vehicular network,a resource allocation scheme of mm Wave vehicular network oriented to network energy efficiency optimization is proposed by considering base station power adjustment and MAB model.This scheme allows the transmission power of mm Wave base station to be adjusted to improve energy efficiency.Therefore,power consumption can be reduced as much as possible on the premise of meeting the basic data rate requirements of vehicle users.At the same time,the sharing of mm Wave can save communication resources and reduce the occupation of radio frequency chain by assigning vehicle users requesting the same data content to the same receiving group.The feasibility of the honeycomb aided mm Wave vehicular communication architecture and the convergence of the proposed model are analyzed theoretically.Simulation results show that the proposed scheme can significantly improve energy efficiency in different scenarios compared with existing schemes.