Adjustable threshold LAS massive MIMO detection under imperfect CSI and spatial correlation

Abstract

In this paper, a likelihood ascent search (LAS) detector with adjustable threshold (ρ-LAS) is proposed for uplink (UL) massive multiple-input-multiple-output (M-MIMO) systems. The performance-complexity tradeoff for the ρ-LAS-based detector is extensively analyzed and compared with the conventional LAS M-MIMO detector by means of Monte Carlo simulations (MCS). Adjusting a threshold associated with the likelihood function, for each system and channel scenario, we found that the ρ-LAS is able to achieve better performance than the conventional LAS detector without complexity increment. Considering practical scenarios deteriorated by antenna correlation and imperfect channel state information (CSI) in M-MIMO systems, ρ-LAS detector has proven to be superior than the LAS detector in terms of performance while requires a fixed but very marginal additional number of computations. In addition, ρ-LAS provided a much better performance-complexity tradeoff in scenario with medium signal-to-noise ratio (SNR) and high number of antennas, a common operation scenario in M-MIMO systems. Finally, the ρ-LAS M-MIMO and three representative M-MIMO detection methods, namely the polynomial expansion (PE), the dual band Newton inversion (DBNI) and the iterative sequential detector (ISD), are compared. The results indicate a substantial performance-complexity tradeoff improvement for our proposed ρ-LAS detector.