Research On Channel Estimation Algorithm In Massive MIMO-OFDM Systems Based On Compressed Sensing

Massive MIMO(massive multiple-input multiple-output,massive MIMO)technology uses a large-scale antenna array on the base station side to make use of airspace resources to make up for the insufficiency of time-frequency domain resources and provide greater system capacity and higher energy efficiency,and supports multi-user connections.Orthogonal frequency division multiplexing(OFDM)technology uses orthogonal subcarriers to transmit data in parallel.It can not only flexibly load according to channel characteristics,improve spectral efficiency,but also suppress inter-symbol interference(ISI)in order to overcome frequency selective fading.In massive MIMOOFDM systems,in order to perform reception detection in uplink(UL)and transmission precoding in downlink(DL),accurate channel state information(CSI)at base stations is required.Therefore,the key to fully exploiting the advantages of massive MIMO-OFDM technology lies in effective channel estimation.Massive MIMO-OFDM wireless channels have sparsity in the beam domain and time delay domain.Channel estimation based on compressed sensing theory utilizes the sparsity of the channel,which can not only improve the channel estimation accuracy,but also reduce the overhead of pilot.This thesis focuses on the channel estimation techniques for large-scale MIMO-OFDM systems.It utilizes the sparsity of channels in the beam domain and time delay domain to estimate the channel based on the theory of compressive sensing.The main research contents summarizes as follows:(1)With the aid of compressive sensing theory,a channel estimation algorithm for large-scale MIMO-OFDM systems based on pilot optimization distribution is studied.In massive MIMO-OFDM system,the pilot distribution is closely related to the channel estimation performance.We can improve the performance of channel estimation by optimizing the pilot distribution.First,the channel estimation problem based on compressive sensing transforms into a block sparse signal reconstruction problem.Then the error boundary is derived by using block sparse signal reconstruction to obtain the relationship between the mutual coherence of all the measurement matrices and the reconstruction performance of the BOMP algorithm.Next,according to the proposed criterion,the GA-based method is presented to obtain the optimized pilot locations.The simulation results show that the proposed method for minimizing the complete block correlation value significantly improve the channel estimation performance compared with the unoptimized pilot algorithm.(2)Based on the block sparsity of multi-user massive MIMO systems,a block estimation algorithm is studied.Firstly,we analyze the spatial correlation and temporal correlation of the channel in multi-user massive MIMO-OFDM system,and then the block sparsity of the channel is derived.Then the channel estimation problem transforms into a block sparse signal reconstruction problem,and a block sparse B-CoSaMP algorithm is proposed to reduce the pilot overhead required for channel estimation.The results show that the B-CoSaMP algorithm has a better estimation performance,and has a great improvement in channel estimation accuracy and bit error rate.