Low-complexity High-performance Receiver Design For Massive MIMO Systems

Massive multiple-input multiple-output(M-MIMO)is considered as one of the key technology in the fifth generation(5G)wireless communications due to its significant improvements in terms of device connection,spectrum efficienty,and converage range.In an M-MIMO system with large number of antennas and users,the realization of receiver with low computational complexity and approximate maximum likelihood(ML)detection performance is the premise of the practical application of the M-MIMO technology.To address this issue,an in-depth study on the low-complexity high-performance receiver design for M-MIMO systems is carried out in this thesis.The main research contents are summarized as follows:1)Aiming at the significantly high computational complexity of the receiver based on the expectation propagation(EP)algorithm for M-MIMO systems,a low complexity receiver design method is proposed.According to the properties of message passing and channel decoding,a joint iterative detection and decoding(IDD)receiver design based on the EP algorithm is presented.To reduce the computational complexity,a novel location-aware and reliability-aware based EP(LR-EP)algorithm is proposed.The constellation location of the symbol to be estimated is determined by the low-complexity quadrant detection and coordinate migration,and the reliabilities of nodes are determined by the predefined reliable-probability threshold.As a result,the calculation of posterior information and message passing is significantly reduced.Simulation results illustrate that the proposed LR-EP based receiver achieves substantial computational complexity reduction compared with the conventional EP receiver with negligible performance degradation.2)Aiming at the deterioration of convergence and robustness of the EP detection caused by the loops in the factor graph for a fully connected M-MIMO system,an information-lossless sparsification transformation(ST)framework is proposed.By utilizing the ST framework,a large portion of loops in the factor graph are eliminated and the degree of nodes is reduced.As a result,the multi-user interference is suppressed,and the convergence and detection performance are enhanced.Based on the state evolution(SE)analysis,the efficiency of the proposed scheme is proved theoretically.Considering the impulsive noise scenario,a robust receiver design based on the S-EP(ST-based EP)algorithm is proposed.The detrimental impulsive noise component is dispersed to multiple physical resource elements,thus mitigating the adverse effects of impulsive noise.The resulting sparsified factor graph alleviates the adverse effects of propagated non-Gaussian messages caused by impulsive noise during the iterative process.Therefore,the robustness of the receiver is enhanced.Simulation results demonstrate that the proposed S-EP based receiver significantly improves the performance compared with the EP based one,especially for the system with a high overload-ratio,high-order modulation,and high probability of impulsive noise occurrence.3)Aiming at the constraints of baseband data bandwidth and chip input/output(I/O)interface for the M-MIMO system with centralized processing architecture,an efficient receiver design based on fully decentralized(FD)architecture is proposed.The S-EP based FD receiver is designed to enhance the detection performance and reduce the iterative computational complexity by utilizing the proposed ST scheme.In order to mitigate the performance deterioration caused by the constrained information-sharing between antenna clusters,a variance compensation scheme is proposed to improve the accuracy of message passing,thus leading to the performance enhancement.Based on the properties of message passing,an efficient non-linear fusion scheme is proposed.Theoretical analysis demonstrates that the proposed non-linear fusion scheme is superior to the conventional linear one.By utilizing the assumption of large system limit and approximation theory,the performance bound and convergence values of the FD receiver are derivated.Simulation results illustrate that the proposed receiver exhibits superior performance-complexity trade-off compared with the existing counterparts.