MIMO Maximum Likelihood Soft Demodulation Based on Dimension Reduction

Abstract

This paper proposes a dimension reduction maximum likelihood (ML) soft demodulator for multiple-input multiple-output (MIMO) system. The proposed dimension reduction soft demodulator reduces the dimension of the search space for minimum Euclidean distance calculation by exploiting ML hard detection algorithms and can achieve significantly lower complexity than existing optimal soft demodulators based on exhaustive or near-exhaustive search. The dimension reduction can be realized through the use of optimum MIMO hard detectors such as a sphere decoder, resulting in optimal performance. It can also be implemented using suboptimum low-complexity MIMO hard detectors such as various conventional MIMO equalizers, further reducing the complexity with slight performance degradation.