Abstract
The maximum likelihood (ML) detection for multiple-input multiple-output (MIMO) system achieves the optimal performances at the cost of high computational complexities, while the linear detectors attain low complexities with degraded performances. In this paper, we propose two low-complexity detection schemes based on the minimum-mean-square-error (MMSE) detection for MIMO systems. Using on the MMSE-detected data as the starting point, the first scheme searches the constellation subspace using the ML criterion. To further reduce the complexity, the second scheme selects the constellation subspace for search using the reformulated ML criterion. Simulation results show the substantial performance improvements compared to the MMSE detection, with only slightly increased complexities. ©2010 IEEE.