Multiple measurement vector based Bayesian learning for simultaneously sparse time/delay-domain channel estimation in ADO-OFDM visible light systems

Abstract

A multipath channel impulse response (CIR) estimator is proposed by leveraging the simultaneous sparsity inherent in the multipath CIR across multiple measurement vectors (MMV) for asymmetrically clipped direct current-biased optical OFDM (ADO-OFDM) visible light communication (VLC) systems. A comprehensive multipath CIR model is developed to account for both specular and diffusive reflections encountered in the VLC propagation environment. We begin by formulating the system model of the ADO-OFDM-VLC system. Following this, we briefly revisit the traditional linear minimum mean square error (LMMSE) and least squares (LS) channel estimators, along with the class of compressive sensing (CS)-based channel estimation (CE) schemes. Specifically, the FOCal Underdetermined System Solver (FOCUSS), its MMV-based extension (MFOCUSS), and orthogonal matching pursuit (OMP) algorithms are considered, as they effectively exploit the sparsity structure present in the multipath CIR of VLC channels. Furthermore, we introduce an enhanced estimation technique—namely, the simultaneous sparse OMP (SOMP)—which effectively utilizes the simultaneous sparsity observed in the delay-domain CIR across MMVs, particularly relevant to the non-line-of-sight (NLoS) components of the VLC channel. In addition, an advanced MMV-based Bayesian learning (MBL) framework is proposed to further reduce pilot overhead by exploiting both time and delay-domain sparsity of the CIR. For benchmarking, the Oracle-based minimum mean square error (O-MMSE), Oracle-based LS (O-LS), and the Bayesian Cramér-Rao lower bound (BCRLB) are utilized. Simulation results confirm that the proposed MMV-based MBL approach significantly outperforms conventional LS, LMMSE, and existing CS-based techniques, including OMP, SOMP, FOCUSS, MFOCUSS, and Bayesian learning (BL) methods, in terms of normalized mean square error (NMSE), pilot overhead, bit error rate (BER), and outage probability (OP).

More information

Identifiers

ORCID for Lajos Hanzo: orcid.org/0000-0002-2636-5214

Catalogue record

Export record

Altmetrics