Research On Channel Estimation And Signal Detection Algorithms For Massive MIMO Systems

Due to the huge improvements in spectral efficiency and energy efficiency,massive multiple-input multiple-output(MIMO)has become one of the key technologies in the 5th generation mobile communication systems(5G).However,massive MIMO system brings many new problems while increasing the system capacity.These problems include antenna array design,preprocessing and beamforming algorithms,pilots design,complexity and performance problems of channel estimation and signal detection algorithm.Among them,channel estimation and signal detection are important modules at the receiving end of massive MIMO systems,which have a huge impact on system performance and time delay.Traditional MIMO channel estimation and signal detection algorithms cannot be applied to massive MIMO systems due to their high complexity or insufficient performance.Aiming at the above problems,this paper separately studies channel estimation and signal detection algorithms suitable for massive MIMO,and proposes optimized algorithms.The main research contents are as follows:(1)The Deep Learning-Based Channel Estimation(DCE)algorithm based on deep learning is analyzed,and a Total Variation DCE(DCE-TV)algorithm is proposed to optimize the cost function.The DCE algorithm first denoises the received signal,and then performs a least squares estimation.The loss function is an implicit squared loss function.Simulations show that the performance of the DCE algorithm is still far from the optimal channel estimation algorithm.Adding an explicit Total Variation regularization term to the function can improve the performance of the DCE algorithm.Simulation result shows that the complexity of the proposed algorithm is slightly increased compared to the original DCE algorithm,and the performance is significantly improved.(2)The analysis focuses on likelihood ascent search(LAS)detection algorithm and 'reactive tabu search(RTS)detection algorithm based on neighborhood search,and proposes a symmetric successive over-relaxation(SSOR-RTS)altorithm that optimizes the initial solution.Simulations show that the performance of SSOR-RTS algorithm is close to that of traditional RTS algorithm,and the complexity is greatly reduced.Based on the SSOR-RTS algorithm,this paper further proposes an adaptive SSOR-RTS algorithm that uses the maximum likelihood(ML)cost change rate to control the number of SSOR iterations.SSOR-RTS algorithm gets a better balance between performance and complexity.Simulation result shows that the performance of the adaptive SSOR-RTS algorithm is equivalent to that of the traditional RTS algorithm,and the average complexity is lower than that of SSOR-RTS.