Research On Detection Algorithm For Uplink Massive MIMO Systems

With the increase of the number of antennas,the complexity of the uplink signal detection algorithm for Massive MIMO systems increases significantly,the original detection algorithm cannot be applied in reality.Therefore,the focus of this paper is the signal detection algorithm with low complexity and good performance under Massive MIMO systems.Firstly,this paper focuses on two kinds of Massive MIMO signal detection algorithms:Likelihood Ascend Search?LAS?algorithm and Reactive Tabu Search?RTS?algorithm.The RTS algorithm can obtain the global optimal solution vector based on the"escape"strategy,therefore the performance is better than that of LAS algorithm,but the complexity is higher than that of LAS algorithm.With the increase of the number of transceivers,the initial calculation complexity occupies an increasing proportion of the overall algorithm complexity,therefore the reduction of the initial calculation complexity is also important for the overall complexity of the RTS algorithm.Secondly,an improved RTS detection algorithm based on Gauss-Seidel?GS?iteration method is proposed,which is to replace the original MMSE algorithm by GS iterative method.The simulation results show that when the number of transmit and receive antennas is 128×16 and the SNR is 16dB,compared with the original RTS algorithm,the complexity of the initial solution vector calculation of the improved algorithm is reduced by 78.5%with 1.6×10^-5 bit error rate performance loss in 4QAM modulation.However,with the increase of modulation order,the gap of two algorithms BER performance is enlarged,so the GS-RTS algorithm is proposed to improve the BER performance in the high order modulation.Finally,this thesis introduces the block constellation?BC?method in GS-RTS algorithm for high order modulation.The block constellation method is used to compute the initial solution vector of GS iterative method,which can accelerate the convergence of GS iterative method and improve the performance of GS algorithm,to further improve the overall performance of GS-RTS algorithm.The simulation results show that when the number of transceivers is 128×16 and the SNR is 16dB,compared with the GS-RTS algorithm,the complexity of the improved BC-GS-RTS algorithm is increased by 12.5%,but the BER performance of the BC-GS-RTS algorithm is improved by 3 times and can be approximated to the orginal RTS algorithm in 16QAM modulation.Therefore,the improved BC-GS-RTS detection algorithm can well replace the original RTS signal detection algorithm in the high order modulation.