Researches On Simplified Matrix Inversion Based Low-complexity Massive MIMO System Signal Detection Algorithm

As an extension of MIMO technology,massive MIMO has significantly improved the channel capacity,spectral efficiency,transmission rate and reliability of wireless communication systems.However,with the number of antennas increasing,reducing the computational complexity becomes an urgent problem to be solved in massive MIMO signal detection.Because of the asymptotic orthogonal between massive MIMO channels,the linear detection algorithms can achieve near optical performance with lower computational complexity.However,the MMSE detection algorithm involves complicated matrix inversion,thus making it difficult to implement rapidly and effectively.Aiming to solve the problem,this thesis first proposes a simplified approximate message passing(SAMP)method to reduce the complexity of matrix inversion.Then a few simplified algorithms based on matrix inversion reduction are introduced comprehensively.These algorithms,namely,Neumann series expansion,multistage linear receiver(MLR),Newton iterative approximation,Richardson iterative algorithm,Jacobi iterative algorithm,successive over-relaxation iterative algorithm,symmetric successive over-relaxation iterative algorithm(SSOR),conjugate gradient(CG)iterative algorithm and SAMP algorithm,are divided into three types of simplified algorithm,with matrix inversion approximation,linear equation solving and approximation information passing respectively.Furthermore,according to the different characteristics of each simplified algorithms,the corresponding optimization schemes are obtained,such as the proper initial value,the proper relaxation parameter and the effective noise variance.Finally,the performance and computational complexity of each type of algorithms are verified by simulation results.Among simplified algorithms based on matrix inversion approximation,MLR algorithm is optimal;Among simplified algorithms based on linear equation solving,SSOR iterative algorithm are the most desirable algorithms;As for the SAMP method,it achieves best trade-off between performance and computational complexity in the all simplified algorithms.Secondly,this thesis also proposes to apply the simplified algorithms based on matrix inversion reduction to MMSE soft interference elimination method for massive MIMO systems,which eliminates the interference of the received signal by employing the soft symbol estimation to further improve the signal detection performance.Moreover,the calculation scheme of effective posteriori LLR and the self-iterative scheme are given to optimize the overall performance of the algorithm.According to the simulation results,it is illustrated that the signal detection performance based on the matrix inverse reduction algorithm in the MMSE soft interference cancellation scheme is significantly improved compared with that of the hard decision scheme.Considering the processing capability of the base station,even although the complexity is very high,it is still a candidate scheme for practical application scenarios.