A Clustering Detector for Spatial Modulation System

摘要

Spatial modulation (SM) is regarded as a promising multiple-input multiple-output (MIMO) technique since it exploits the additional spatial domain for transmitting information. For transceiver design in SM, the traditional schemes first estimate the channel state information (CSI) based on pilots for further decoding, which introduce significant overhead, thereby reducing the spectral efficiency of the transmission. In this paper, to avoid the channel estimation, we convert the signal detection problem into a cluster problem for SM system. Utilizing the inter-structure of the SM system, we propose a structured Gaussian mixture model (SGMM) with a reduced number of parameters to model the received signals. We also develop an SGMM-based expectation maximization algorithm (SGMM-EM) to cluster the received signals. The complexity of SGMM-EM algorithm is reduced significantly compared with traditional clustering algorithm. After clustering all the received signals, a label mapping method is developed to map each transmitted symbol to one of the clusters. Simulation results show that, with only a small number of iterations, the performance of our proposed scheme approaches to that of the optimal receiver with perfect CSI.