Research On Signal Detection For Massive MIMO Systems Via ADMM Approaches

Massive multiple-input multiple-output（MIMO）is considered as one of the most important techniques in the 5^th and 6^th generation（5G and 6G）of communication systems.Among multiple technologies of massive MIMO,signal detection at the receiver has huge impact on total system performance.However,as the number of antennas increases gradually in practical communication system,the complexity of signal detection would dramatically increase correspondingly.In applications with hundreds of antennas in massive MIMO system,classical signal detection methods generally have a bottleneck problem,that is to make trade-off between detection performance,complexity,and signal processing parallelism.With the development and inter-crossing of various disciplines,optimization theory is widely used to deal with practical problems in different fields.Due to simplicity,operator splitting capability and theoretically-guaranteed performance,the alternating direction method of multipliers（ADMM）is widely used to solve large-scale convex and nonconvex problems.In this thesis,for quadrature amplitude modulation（QAM）systems,we exploit ADMM technique combined with several improved maximum likelihood（ML）detection models and propose several novel massive MIMO detection methods,which have excellent detection performance and low computational complexity.The major contributions of this thesis are summarized as follows.1.Aiming at the problem that the state-of-the-art massive MIMO detection algorithms cannot offer attractive trade-off between detection performance and complexity,we design an efficient QAM signal detection method for massive MIMO communication systems via the penalty-sharing ADMM（PS-ADMM）technique.First,we transform the ML detection model to a nonconvex sharing optimization problem for massive MIMO-QAM systems.Here,a high-order QAM constellation is decomposed to a sum of multiple binary variables,the integer constraints are relaxed to box constraints,and the involved quadratic penalty functions are added to the objective function to pursue a favorable integer solution.Second,a customized algorithm,named PS-ADMM,is proposed to solve the formulated nonconvex optimization problem.Third,performance analysis of the proposed PS-ADMM algorithm,including convergence property and computational cost,are provided.In the end,simulation results demonstrate the effectiveness of the proposed approach in comparison with state-of-the-art approaches.2.To overcome the difficulty that PS-ADMM has to select proper nonconvex penalty parameters,we proposed two signal detection methods,named?₂-box ADMM and Simple?₂-box ADMM（S?₂-box ADMM）,for massive MIMO-QAM systems.First,we develop an?₂-box ML formulation for massive MIMO-QAM signal detection and customize the ADMM algorithm to solve the nonconvex optimization model.Second,in order to reduce the computational complexity of?₂-box ADMM and improve its detection performance,the S?₂-box ADMM algorithm is proposed.Third,several theoretical results related to convergence,iteration complexity,and computational complexity are presented.Simulation results verify that the improved S?₂-box ADMM algorithm is comparable to the PS-ADMM algorithm in terms of BER detection performance and computational complexity.3.In order to improve the detection efficiency of PS/?₂-box/S?₂-box ADMM algorithm,we customize the proposed PS/?₂-box/S?₂-box ADMM algorithms to the distributed ones,named DPS/D?₂-box/DS?₂-box ADMM.In the implementation,all variables in proposed distributed algorithms are solved analytically and in parallel,which improve signal detection speed of the proposed PS/?₂-box/S?₂-box ADMM algorithms.In summary,facing the difficulty of high-order QAM signal detection in mobile communication system beyond 5G or 6G,this dissertation proposes two detection algorithms and their variants based on the ADMM technique with a good balance between excellent detection performance and low computational complexity.The proposed algorithms have the potential ability to Uplink Centric Broadband Communication（UCBC）scenarios of 5.5G mobile communication system,which provides new ideas and methods for the theoretical research and engineering practice in the field of wireless communications.