Low-complexity algorithms for large-MIMO detection

Low-complexity near-optimal detection of large multiple-input multiple-output (MIMO) signals has attracted recent research attention. Recently, it has been shown that certain algorithms rooted in machine learning/artificial intelligence are well suited to achieve near-optimal detection performance in MIMO systems with large number (tens) of antennas at practically affordable complexities. In this paper, we present three such low-complexity algorithms that we have proposed recently, and compare their bit error rate performance and complexities in large-MIMO detection. These algorithms include two local neighborhood search based algorithms, namely, likelihood ascent search (LAS) and reactive tabu search (RTS) algorithms, and a message passing algorithm based on and belief propagation (BP). Feasibility of such low-complexity algorithms for large-MIMO detection can enable practical implementation of high-spectral efficiency (tens to hundreds of bps/Hz) large-MIMO systems.