Research On Millimeter Wave Massive MIMO Hybrid Precoding Algorithm

With the continuous development and wide application of wireless communication technology,wireless data services are exploding,and the use of millimeter waves for wireless communication has become a fundamental way to solve the shortage of spectrum resources and increase system capacity.In order to resist severe path fading of millimeter wave signals,a large-scale MIMO structure must be used to improve the signal transmission and reception gain.However,for large-scale MIMO antenna arrays,it is difficult for the system to provide independent RF links and their associated hardware circuits for each antenna,and it is difficult to obtain complete channel state information at the signal transmitter.Therefore,it is necessary to estimate the statistical characteristics of the channel effectively,and propose a low complexity hybrid precoding strategy suitable for millimeter wave systems.Focusing response to this problem,this paper mainly completes the following work:The principle of continuous basis pursuit(CBP)is used to construct the beaming dictionary matrix.In order to ensure the efficient operation of the adaptive estimation process,And a multi-resolution codebook is designed to reduce the complexity of channel parameter estimation in millimeter-wave massive MIMO systems.At the same time,the channel state information reconstruction is completed by using the estimated channel parameters,and the performance of the hybrid precoding is evaluated by the SOMP hybrid precoding algorithm.Experiments show that the beaming dictionary based on continuous basis pursuit can improve channel estimation accuracy.(2)The channel estimation method based on the uniform grid dictionary obtains the beaming space channel containing the lower dimension of the path gain information,and performs singular value decomposition on the beaming space channel matrix to obtain the optimal precoding matrix.Then,in order to reduce the computational complexity of hybrid precoding,a discrete Fourier transform-mutual unbiased base(DFT-MUB)codebook is designed to apply mutual unbiased base codebook to digital beamforming,and discrete Fourier codebook used in RF analog beamforming.At the same time,the analog precoding and digital precoding design is completed by the beaming space hybrid precoding algorithm(BHP),which avoids a large number of "matrix inversion" in the conventional iterative algorithm.Compared with the all-digital precoding method,not only the performance loss is less than 5%,but also the computational complexity is reduced.