LAS Detector with Soft-Output MMSE Initialization Under Imperfect Channel Estimation and Channel Correlation

• Published in: Wireless personal communications Vol. 108; no. 1; pp. 213 - 220
• Main Authors: Chihaoui, Issa, Ammari, Mohamed Lassaad
• Format: Journal Article
• Language: English
• Published: New York Springer US 15.09.2019; Springer Nature B.V
• Subjects: Algorithms; Ascent; Communications Engineering; Computer Communication Networks; Computer simulation; Correlation; Decoding; Engineering; Error detection; Maximum likelihood decoding; MIMO (control systems); Networks; Sensors; Signal,Image and Speech Processing; LAS algorithm; MIMO; Imperfect channel estimation; MMSE filter; Correlated channel; ML detector
• ISSN: 0929-6212; 1572-834X
• DOI: 10.1007/s11277-019-06397-9

This paper proposes a modified likelihood ascent search (LAS) algorithm for multiple-input multiple-output (MIMO) systems under correlated channel and imperfect channel state information. We assume that the LAS detector is initialized by the minimum mean square error (MMSE) filter. We propose a new maximum-likelihood (ML) decoding metric by considering the soft-output of the MMSE filter. Our proposed LAS takes both channel estimation errors and spatial correlation of antennas into account when computing the ML metric. We show that, while increasing the computational complexity, the proposed LAS remains suitable for massive MIMO systems. Simulation results show that the proposed LAS outperforms the conventional LAS detector, specially when the transmitter correlation increases.