Based On Tensor Decomposition Channel Estimation Algorithm For Millimeter Wave Massive MIMO System

Millimeter wave massive MIMO technology is an important technology in the fifth generation mobile communication system.The energy efficiency,spectrum efficiency and communication capacity of the system can be improved by installing a large scale antennas array at the transmitter and receiver.With the increasing of the number of receiving and transmitting antennas,However,the communication system becomes more and more complex,and how to estimate accurately the channel state information has become a hot topic in the study of mm Wave MIMO systems.Aiming at this problem,this paper makes full use of the multidimensional characteristics of signals and combines with the processing capacity of tensors for large-scale data,conducts tensor modeling of large-scale received signals at the receiving end,and studies the following research on the channel estimation problem of millimeter wave massive MIMO:Firstly,in order to solve the problem of spatial multi-dimensional beam interference in mm W massive MIMO system,a mm W massive MIMO enhanced channel estimation algorithm based on non-negative tensor decomposition is proposed.With space blanking converts multidimensional space beam interference signals to non-negative signals to be enhanced,then,the method of using nonnegative tensor decomposition tensor decomposition of signal to get accurate channel estimation,the simulation results show that the performance compared with other algorithms,the proposed channel estimation algorithm can effectively reduce the BER of system.Secondly,in order to solve the problem that the traditional tensor decomposition is difficult to estimate the high-dimensional channel due to the increase of antennas array in mm W massive MIMO system,a channel parameter estimation method based on grid-parafac mm W massive MIMO system is presented.By using grid-parafac,the large-scale received signal tensor is first converted into several sub-received signal tensors,then the parallel tensor decomposition is carried out,and the parallel iteration of the ALS algorithm is derived to accurately obtain the channel parameter factor matrix.Simulation results show that the proposed algorithm can effectively reduce the computational complexity.Finally,in multiuser hybrid precoding mm W massive MIMO system,with the increase of the number of receiving and sending antennas,the channel transmission matrix needs higher dimensional operation.An algorithm of dimension reduction channel estimation based on FCP is proposed.The low-dimensional embedding of multi-user received signals by FCP dimensionality reduction algorithm can accurately estimate the channel factor matrix in the form of Kruskal,and maintain the channel characteristics of the original received signals while dimensionality reduction of multi-user received signals.Simulation results show that the FCP reduced dimension channel estimation algorithm can effectively improve the system performance.