A Unified Model for Signal Detection in Massive MIMO System and Its Application

Linear minimum mean-square error (MMSE) detection method bears high computational complexity due to matrix inversion operations in massive multiple-input multiple-output (MEMO) systems. Existing methods to address this problem can be divided into approximation methods and iterative methods. In this paper, we introduce the relationship between the two types of methods. Furthermore, a simple approach is developed to determine the appropriate relaxation parameter for the symmetrical successive over relaxation (SSOR) method. Finally, simulation results verify the proposed relationship and demonstrate the superiority of the proposed relaxation parameter.