Research On Channel Estimation Technology In A Massive MIMO System

Massive MIMO is one of the most important technologies of 5G.By configuring a large number of antennas(100-1000)in the base station,the energy can be concentrated in a small space to improve throughput.At the same time,the antenna makes full use of the space freedom,using the same time-frequency resources and multiple terminals(16-64)simultaneous communication to reduce overall energy efficiency.However,the premise of using these excellent characteristics is that the base station must estimate the downlink channel,especially when the number of antennas increases,the traditional channel estimation scheme will bring huge pilot overhead,and channel estimation becomes almost impossible.This thesis mainly discusses the downlink non-blind channel estimation algorithms under massive MIMO.The main contents are as follows:Firstly,an improved algorithm for structured compressive sensing that can adapt to sparsity is proposed.Objectively,wireless channels are sparse.Subspace pursuit algorithms can accurately estimate multiple channels from a large number of transmitting antennas at the base station by introducing a backtracking mechanism.Aiming at the algorithm's need for channel sparsity as a prior condition,this thesis uses a non-orthogonal pilot structure combined with the space-time correlation of wireless channels,and proposes an adaptive matching pursuit algorithm with block sparsity.Simulation results show that the proposed algorithm is not only equivalent in performance to the subspace pursuit algorithm,but also can accurately estimate the sparseness of the channel.Secondly,based on the channel spatial-temporal correlation,an improved convex optimization algorithm is proposed.The iterative support detection algorithm can reconstruct sparse signals with fast fading characteristics in amplitude with a relatively low pilot overhead.This thesis proposes a structured iterative support set detection algorithm based on the iterative support detection algorithm combined with the space-time correlation of the channel.It firstly solves the base tracking problem through an alternating direction algorithm,and then updates the initial signal through a threshold detection strategy support set.Finally,the final support set is determined based on the structure of the signal.After the three processes are performed alternately several times,the sparse channel matrix estimate can be obtained.Simulation results show that the NMSE curve of the proposed algorithm is lower than that of the original algorithm,and it has a maximum gain of about 6d B in a low SNR environment.Finally,a feedback algorithm based on differential operation is proposed to reduce the feedback overhead.The fading channel is essentially a stochastic process with chromatograms,which means that there is redundant information between adjacent symbols.The proposed algorithm firstly uses compressed sensing technology.The downlink channel matrix is estimated to be obtained at the user terminal,and then a differential operation is performed on any two adjacent symbols to reduce redundancy to obtain the channel matrix differential value.Finally,the differential value is fed back to the base station as a signal compression through the uplink.Simulation results show that the feedback algorithm based on differential operation has higher reconstruction accuracy under the same number of pilots than when it is not used,indicating that this scheme can further reduce the feedback overhead.