Research On Sparse Channel Estimation Method For MIMO System Based On Compressed Sensing

Multiple input multiple output(MIMO)systems can significantly improve the channel capacity and transmission rate by deploying multiple antenna arrays at the transmitter and receiver,which is considered as one of the key enabling technologies of wireless communication in the future.However,with the increase of the number of antennas at the base station in MIMO systems,a large number of unknown channel parameters and huge pilot overhead are generated,which makes wireless channel estimation a challenging problem.Therefore,the research on channel estimation of MIMO systems is of great significance to the development of wireless communication technology in the future.In wireless communication systems,the wireless multipath channel tends to exhibit sparse characteristics,and the traditional channel estimation algorithms do not take advantage of the sparsity of the channel.The channel estimation methods based on compressed sensing(CS)can make full use of the sparse characteristics of the channel,and use few pilot sequences to obtain better channel estimation performance.In this paper,a series of sparse channel estimation methods based on compressed sensing are proposed for MIMO-OFDM(multi-input multi-output orthogonal frequency division multiplexing)and massive MIMO systems in complex environments.The main contents of this paper are as follows.1.First of all,an improved sparse channel estimation algorithm for MIMO-OFDM systems based on CS is proposed.Based on the wireless channel model of MIMO-OFDM systems,a new adaptive matching pursuit(NAMP)sparse channel estimation algorithm is proposed to solve the problem of unstable reconstruction performance of the existing sparse-based adaptive matching pursuit channel estimation algorithms under low signal to noise ratio.This method can improve the convergence efficiency by selecting the support set atoms in the iterative process through the fixed step size without prior knowledge of the channel sparsity.Furthermore,the mechanism of singular entropy order determination is adopted to prevent the introduction of irrelevant atoms and improve the convergence accuracy of the algorithm.Experimental results show that the proposed NAMP method has less computational complexity and more stable performance.2.Moreover,an improved sparse channel estimation algorithm for time division duplex(TDD)massive MIMO systems based on CS is proposed.This paper discusses the pilot pollution and proposes an effective semi orthogonal pilot design scheme for the uplink model of TDD massive MIMO systems.Furthermore,considering that the existing algorithms are not sensitive to the tap energy and the reconstruction precision is not high,an optimized adaptive matching pursuit(OAMP)algorithm is proposed.The energy entropy-based sorting method is adopted to screen the support set atoms and improve the estimation performance of the algorithm.Then,a piecewise adaptive variable step size method is utilized to improve the generalization ability of the algorithm.The theoretical analysis and simulation results show that the proposed OAMP algorithm further reduces the pilot pollution and improves the accuracy of channel estimation at the cost of less time complexity,its comprehensive performance is better than other channel estimation algorithms.3.Then,an improved sparse channel estimation algorithm for frequency division duplex(FDD)massive MIMO systems based on structured compressed sensing is proposed.Aiming at the problem that it is difficult to determine the channel sparsity and the threshold parameters of the reconstruction algorithm in the downlink model of FDD massive MIMO systems,this paper proposes a time-frequency block sparse channel estimation method based on structured compressed sensing,which is named as generalized block adaptive matching pursuit(g BAMP)algorithm.The proposed g BAMP algorithm uses the time-frequency joint block sparsity of massive MIMO systems to optimize the selection of index set,so as to improve the stability of the algorithm.Then,in the absence of the threshold parameter,the algorithm uses the F-norm of the residual to determine the stop condition of the adaptive iteration,and proves the effectiveness of the algorithm.Simulation results show that the proposed g BAMP algorithm can quickly and accurately estimate the channel state information of FDD massive MIMO systems,and the performance of this method is better than other similar algorithms.4.Finally,a deep learning based sparse channel estimation algorithm for compressed sensing FDD massive MIMO systems is proposed.In FDD massive MIMO systems,with the increase of the scale of wireless channel matrix,the disadvantages of intensive computing of the iterative optimization and unable to guarantee the global optimal solution have become the bottleneck of the application of CS in channel estimation.In order to solve the problem,this paper proposes a new method of deep learning for CS-based sparse channel estimation in FDD massive MIMO systems,which is named as convolutional compressive sensing network(Con CSNet).In the case that the channel sparsity is not required,this paper uses the Con CSNet algorithm to solve the inverse transformation process of channel state information obtained from the received signal,so as to solve the underdetermined optimization problem under the framework of CS and realize the reconstruction of the original sparse channel.The experimental results show that the proposed Con CSNet algorithm has higher accuracy and faster operation speed than the traditional greedy algorithms.