Research On Massive MIMO Channel Estimation Algorithm Based On Sparse Bayesian Learning

Massive MIMO is a promising technique for future 5G communications due to its high spectrum,energy efficiency and link reliability.To realize its potential performance gain,accurate channel estimation is essential.However,due to massive number of antennas at the base station(BS),the pilot overhead increases significantly,which can be effectively reduced by means of channel sparsity.In this thesis,we will utilize the group sparsity of the channel in the delay domain and the hidden sparsity of the virtual angular domain to conduct system models,and use the sparse Bayesian learning(SBL)algorithm as the key algorithm to perform channel estimation.First,in the context of multipath channels for Massive MIMO-OFDM systems,we propose a channel estimation algorithm with joint group sparse and block matching sparse Bayesian learning.It is popularly considered that that the channels between different antennas at the base station side and the same user have common sparsity.The proposed algorithm adopts multipath grouping to enhance the common sparsity,and further reduce the computational complexity.Owing to the large number of blocks in the real scenario,the conventional algorithm is no longer applicable.The block matching Bayesian estimation algorithm uses the proposed block matching algorithm to initially reduce the dimension of the measurement matrix without destroying the block structure.The channel coefficients after dimension reduction are regarded as hidden variables.The sparseness and correlation between each group of channel elements are represented by hyperparameters.The hyperparameters are obtained by the expectation maximization with the expected value representing the channel response.This algorithm can not only improve the estimation accuracy,but also automatically capture the channel sparse information.Secondly,aiming at the channel estimation problem in the angular domain of Massive MIMO uplink systems,this thesis proposes a modified angle mismatch and SBL channel estimation algorithm.By utilizing the hidden sparseness of Massive MIMO in the virtual angle domain,both the samplinggrid of the virtual angle domain and the gaps from the actual direction of arrival(DOA)are considered as adjustable parameters.Moreover,the linear fitting model of angle bias adopts linear interpolation and first-order Taylor expansion,respectively.Multiple parameters are iteratively updated through the sparse Bayesian algorithm,thus obtaining accurate support set,which is further used to estimate the channel coefficients.Compared to the SBL algorithm without angle mismatch,simulations show that the proposed algorithm can achieve more accurate channel state information.