Research On Low Complexity Signal Detection Algorithms In Massive MIMO Systems

The subject of this thesis is supported by the National Natural Science Foundation of China(No.61571108),which is named Research on Low Complexity Key Algorithms in Massive MIMO System.Massive MIMO has the advantages of high spectral efficiency,high transmission speed and high anti-interference ability,which has become a research hotspot in the field of wireless mobile communication.Signal detection technology is one of the core technologies of massive MIMO.It has an important influence on the anti-interference ability and reliability of the entire system.With the increase of the number of massive MIMO system antennas,the signal detection technology of the traditional small-scale MIMO is applied to a massive MIMO system,which has the problems of high computational complexity and poor detection performance.In order to solve these two basic problems,this thesis focuses on two kinds of low complexity signal detection algorithms,linear detection based on SSOR iteration and a message passing detection algorithm based on factor graphs.The main work of this thesis is as follows:(1)Focusing on the significantly high complexity of MMSE detection in massive MIMO systems,three low complexity linear detection algorithms are studied.First,the applications of Neumann series expansion algorithm and Newton iterative algorithm in the massive MIMO linear detection are analyzed,then,a simplified MMSE receiver method based on SSOR iteration in massive MIMO system is proposed.SSOR iterative method is a low complexity method for solving the approximate solution vectors of linear equations,the prerequisite is that the coefficients matrix must be diagonally dominant and symmetric positive definite,both conditions can be satisfied in the massive MIMO system,and the concrete proofs are given in the thesis.In addition,a simple method for obtaining the optimal relaxation coefficients of SSOR iterative algorithm is presented in this thesis.In addition,a method of directly estimating matrix HHH is introduced,which avoids estimating the matrix H by Channel estimation algorithm,reduces the overall complexity of the detection algorithm.According to results of simulations and complexity analysis,it can be concluded that 1)Neumann series expansion algorithm,Newton iterative algorithm and SSOR iterative algorithm avoid the direct inversion of matrix,and reduce the computational complexity of the detection algorithm;2)the convergence speed of MMSE detection algorithm based on SSOR iterative is faster than that based on Neumann series expansion and Newton iteration,and the detection performance is better;3)A better detection performance can be obtained by using the proposed method to estimation matrix HHH.(2)We study the message passing detection algorithm based on factor graphs in massive MIMO system.First,three kinds of graph models are briefly introduced:Bayesian belief network,Markov random field and factor graph.Then,the application of factor graph based belief propagation algorithm in massive MIMO detection is analyzed in detail.The computational complexity of BP-FAG algorithm is analyzed.In addition,in order to improve the convergence speed of BP-FAG detection algorithm,two acceleration methods are introduced in this thesis:Aitken's square incremental method and damping method.Finally,the application of 4QAM in BP-FAG detection algorithm is analyzed.According to the results of simulations and complexity analysis,it can be concluded that 1)with the increase of the number of transmitting antennas and receiving antennas,BP-FAG detection algorithm tends to SISO model performance;2)BP-FAG detection algorithm has low computational complexity and is suitable for massive MIMO systems;3)Aitken's square incremental method and damping method can effectively improve the convergence speed of BP-FAG detection algorithm and improve system detection performance;4)BP-FAG detection algorithm using 4QAM modulation still has good performance.