Sequential likelihood ascent search detector for massive MIMO systems

TL;DR

This paper proposes an optimized likelihood ascent search (LAS) detector for massive MIMO uplink systems, improving performance with a threshold adjustment that reduces required SNR without increasing complexity.

Contribution

It introduces an optimized threshold in LAS for massive MIMO, achieving better BER performance at lower SNR with no added complexity.

Findings

Optimized threshold improves BER performance.

Proposed LAS reduces SNR requirement by 5 dB.

Effective for 32x32 massive MIMO scenarios.

Abstract

In this paper, we have analyzed the performance-complexity tradeoff of {a selective} likelihood ascent search (LAS) algorithm initialized by a linear detector, such as matched filtering (MF), zero forcing (ZF) and minimum mean square error (MMSE), {and considering an optimization factor $\rho$ from the bit flipping rule}. The scenario is the uplink of a massive MIMO (M-MIMO) system, and the {analysis has }been developed by means of computer simulations. With the increasing number of base station (BS) antennas, the classical detectors become inefficient. Therefore, the LAS is employed for performance-complexity tradeoff improvement. Using an adjustable optimized threshold on the bit flip rule of LAS, much better solutions have been achieved in terms of BER with no further complexity increment, indicating that there is an optimal threshold for each scenario. Considering a $32\times32$ antennas scenario, {the large-scale MIMO system equipped with the proposed LAS detector with} factor $\rho$ = 0.8 {requires} 5 dB less {in terms of SNR than the conventional LAS of the} literature ($\rho$ = 1.0) to achieve the same bit error rate of $10^{-3}$.