Efficient approximate-ML detection for MIMO spatial multiplexing systems by using a 1-D nearest neighbor search

Abstract

It is known that suboptimal (equalization-based and nulling-and-cancelling) detectors for MIMO spatial multiplexing systems cannot exploit all of the available diversity. Motivated by the insight that this behavior is mainly caused by poorly conditioned channel realizations, we propose the line-search detector (LSD) that is robust to poorly conditioned channels. The LSD uses a 1-D nearest neighbor search along the least significant singular vector of the channel matrix. It exhibits near-ML performance and has significantly lower complexity than the sphere-decoding algorithm for ML detection.