Research On Transmission Technology Of Millimeter Wave Multi-antenna System

With the rise of technologies such as the Internet of Things,wireless communication technology,as a key factor to promote social and economic development,needs to meet increasing demands.Compared with the fourth-generation mobile communication technology,the fifth-generation mobile communication technology and future communications need to achieve ultra-high throughput,high data transmission rate,ultra-high bandwidth,ultra-low latency,and high reliability.However,the communication technology of the frequency band below 6GHz has basically reached the limit,and cannot meet the development needs of ”ultrahigh capacity,huge connection,wide application”.In order to tap more spectrum resources and realize high-bandwidth,high-performance communication,millimeter wave technology has received widespread attention in the industry.However,considering the characteristics of large millimeter wave path loss,millimeter wave communication needs to combine multi-antenna technology to effectively improve communication quality.Therefore,the transmission scheme of millimeter wave multi-antenna system is studied.The main contents of this thesis are summarized as follows.Firstly,we summarizes the transmission characteristics of millimeter wave channels and typical traditional beam training / beam tracking algorithms.Firstly,the millimeter wave transmission characteristics are introduced in detail,including: atmospheric fading,rain attenuation,large-scale transmission characteristics,and millimeter wave plane and spherical wave transmission characteristics.Secondly,the three channel modeling methods and channel models are studied;Then,the millimeter wave multiple input multiple output(MIMO)system architecture,including: digital system architecture,analog system architecture and hybrid system architecture is discussed.Finally,a codebook design scheme,classic layered beam training algorithm and traditional beam tracking scheme of the extended Kalman filter are given,and the shortcomings of the hierarchical beam training algorithm and the beam tracking algorithm based on the extended Kalman filter are pointed out,which pointed out the research direction for the follow-up work.Subsequently,we studied the transmission technology of MIMO system under fixed line-of-sight wireless communication scenario,and proposed two multi-panel array antenna design schemes.Firstly,considering the accuracy of the spherical wave channel model,the spherical wave channel model is adopted in the fixed-line-of-sight wireless communication system.In order to achieve a high channel space degree of freedom,channel capacity,and a small channel condition number,the azimuth elevation angle and the small up,down,left,and right offset distances are considered at the receiving end,and the design scheme of the multi-panel array antenna is studied(Scheme 1),and the relationship between antenna spacing,transmission distance,array spacing,number of antennas,offset angle and other parameters is derived.Secondly,the design scheme of the multi-panel array antenna(Scheme 2)when the receiving end has a large left and right offset distance is further proposed,and the relationship between various parameters under the condition of maximizing channel capacity and spatial freedom is given.Then,two transmission schemes,including uniform power distribution scheme and beamforming technology are introduced.Finally,the simulation analysis results are given.The simulation results show that the antenna design scheme has greater robustness to angular rotation without considering the larger offset distance.But for those with large left and right(up and down)offset distances and azimuth(elevation angle),ignoring the case of a larger offset distance will result in a lower channel space degree of freedom,and the second scheme given will achieve a higher channel space degree of freedom.For antenna placement with low signal-to-noise ratio and low spatial freedom,the beamforming transmission scheme performs better than the uniform power distribution scheme.But in the case of a high channel spatial freedom,the uniform power distribution scheme is better than beamforming transmission scheme.Finally,in order to the maximum effective channel transmission rate,combining the traditional beam training algorithm with artificial intelligence,two intelligent beam training algorithms are proposed.Firstly,in order to solve the problem that the performance of the hierarchical beam training algorithm is affected by the number of codebook layers to be retrieved,an intelligent hierarchical beam training algorithm based on a multi-armed bandits is proposed,which uses the number of codebook layers to be retrieved in traditional layered beam training as the action setting,and uses the maximum confidence upper limit strategy to intelligently perceive the optimal layered beam scheme in different environments.Secondly,in order to significantly reduce the beam training overhead and perceive the channel time-varying characteristics of the mobile terminal,a smart beam training algorithm based on deep Q network(DQN)is proposed,which only needs to use the beam selection vector as the state space to achieve a large effective channel transmission rate.The simulation analysis shows that the proposed intelligent beam training algorithm based on DQN is superior to the intelligent hierarchical beam training algorithm based on multi-armed bandits,and both algorithms are superior to the direct search algorithm,the traditional hierarchical beam training algorithm and intelligent beam tracking algorithm based on multi-armed bandits using codebook as action.