Clustering-Based Signal Detection for MIMO Systems

In this paper, we propose a clustering-based signal detection method for multiple-input multiple-output (MIMO) systems. Inspired by the phenomenon that the received signals are divided into clusters within block-fading channels, we establish a system constrained Gaussian mixture model (SCGMM) for signal detection, in which the expectation-maximization (EM) algorithm is exploited to calculate the parameters and the a posteriori probabilities (APPs). The quality of the SCGMM clustering depends heavily on the selection of initial centroids. To obtain reliable initial centroids, we design an initialization algorithm, which can reduce the estimated error of the linear minimum-mean-squared-error (LMMSE) estimator and accelerate convergence of the SCGMM clustering. The simulation results show that the bit-error-rate (BER) performance of the proposed method can approach the performance with perfect channel state information (CSI) in the high signal-to-noise (SNR) regions.