Research On Channel Estimation Algorithm Of Millimeter Wave Massive MIMO

The future mobile communication scenario is the Internet of Everything,which has higher requirements for the data transmission rate and capacity of the wireless communication system.Millimeter wave communication technology has a large bandwidth and can meet the needs of high speed and large capacity communication,but at the same time,millimeter wave has large propagation loss.Massive MIMO technology can bring greater antenna gain and can compensate for the propagation loss of millimeter wave.Therefore,millimeter wave technology and massive MIMO can be used together to better meet communication needs.On the other hand,channel estimation affects the performance of wireless communication system.However,due to the strong uncertainty and unpredictability of the channel in the millimeter wave massive MIMO system,it is difficult to estimate the channel.Firstly,this thesis studies the channel estimation algorithm for the switch network based millimeter wave massive MIMO system.By analyzing the system model and channel model,the unique properties of the channel matrix,including sparsity and low rank,are discussed.Based on the physical properties of the channel matrix in the millimeter wave massive MIMO system,a connection is established between channel estimation and matrix completion theory,and the channel estimation problem is reconstructed into a matrix completion problem,that is,the rank minimization problem of the channel matrix.Secondly,for the rank minimization problem of the channel matrix,this thesis performs channel estimation based on nuclear norm minimization.Since the rank function of the channel matrix is non-convex,the nuclear norm is used to approximate the rank function of the channel matrix,and the channel estimation problem is transformed into the nuclear norm minimization problem of the channel matrix.When solving the nuclear norm minimization problem,the IALM is used,and a channel estimation algorithm based on IALM is designed.According to the simulation results,the channel estimation algorithm based on IALM has higher channel estimation accuracy at high signal noise ratio.Under the environment of 25dB signal noise ratio,its normalized mean square error is reduced by about 12.61dB compared with the singular value threshold algorithm,and about 2.66dB compared with the joint alternating direction method of multipliers.Finally,since the nuclear norm treats all singular values in the channel matrix equally,the rank function of the channel matrix can not be well approximated in practice.In order to solve this problem and further improve the performance of the channel estimation algorithm,this thesis relaxes the rank function of the channel matrix into a TNN,and transforms the channel estimation problem into a TNN minimization problem.In TNN,only a small part of the singular values of the matrix are processed,which can better approximate the rank function of the matrix.For this problem,this thesis studies the ADMM of TNN,and uses TNN-ADMM for channel estimation.Through the simulation analysis,it can’ be seen that the estimation algorithm based on TNN-ADMM selects the rank in TNN as 1.Under the environment of 25dB signal noise ratio,compared with the IALM algorithm,the normalized mean square error is reduced by about 2.72dB,the achieved spectral efficiency is improved by about 20.6%.