Research On Channel Estimation Method Of Millimeter Wave Large-scale MIMO System Based On Sparse Reconstruction

With the rapid development of technologies such as Internet of Vehicles,telemedicine,and holographic image,it has become an irreversible trend for 5G networks with high channel capacity to replace traditional 4G networks.In 5G cellular communication technology,the Millimeter-wave(mm Wave)frequency band has received great attention from academia,business and government due to its huge available bandwidth.In addition,the Massive Multiple Input Multiple Output(MIMO)antenna technology is proposed,which not only solves the need to use complex nonlinear structures in traditional MIMO systems to reduce environmental interference such as noise and attenuation,but also increases the complexity of the spatial domain.It greatly increases the system channel capacity.However,as the number of antennas increases by hundreds or thousands,the conventional training overhead required to obtain accurate channel status information(CSI)is proportional to the size of the base station(BS)and mobile station(MS)antennas.grow proportionally.In addition,as the frequency increases,various attenuation losses of mm Wave channels,such as path loss,make the communication environment more unfavorable.Therefore,this paper solves the problem of channel estimation based on the millimeter-wave massive MIMO communication system.This paper introduces the principle of sparse reconstruction in detail,deduces the sparse Bayesian learning algorithm,introduces the development of channel estimation algorithms at home and abroad,and clarifies the low-rank properties of millimeter-wave channels and their sparse properties in the angle domain;the use of hybrid precoding The millimeter wave channel transceiver system model is constructed;the optimized sparse bass sparse channel estimation algorithm based on generalized approximate message passing and the channel estimation algorithm under impulse noise are introduced and advanced.Aiming at the high computational complexity of channel estimation in the traditional sparse Bayesian algorithm under a two-dimensional sparse dictionary,a Generalized Approximate Message Passing(GAMP)algorithm was proposed to replace the high computational complexity in the sparse Bayesian iteration process.Dimensional matrix inversion process.However,since the GAMP algorithm is limited by the dictionary matrix and needs to converge in a random Gaussian matrix,we use the damped generalized approximate message passing algorithm(DGAMP),which slightly increases the convergence time by adding a damping coefficient.to ensure that the algorithm converges during the SBL iteration.Through the 3-point DFT interpolation algorithm,the grid mismatch problem existing in sparse estimation is solved,and accurate channel estimation is realized.Aiming at the situation that most estimation algorithms fail in the presence of impulse noise in the environment,a super-resolution channel estimation algorithm based on Accelerated Structured Alternative Projections(ASAP)is proposed,which makes full use of the channel matrix.Low-rank properties and sparse properties of impulse noise.Furthermore,to reduce the pilot overhead for channel estimation,we adopt a dual Hankel structure in mm Wave channel estimation.The recovered channel matrix contains only few elements due to the specially designed analog-digital hybrid communication structure.Therefore,we use a 2-dimension(2D)harmonic recovery scheme to obtain accurate channel estimates.This paper uses MATLAB tool to carry out a lot of simulations on the proposed algorithm.The simulation results verify the superiority of the proposed algorithm,that is,the proposed algorithm greatly improves the accuracy of channel estimation.