Home // ICDT 2023, The Eighteenth International Conference on Digital Telecommunications // View article
Authors:
Ionut-Dorinel Ficiu
Cristian-Lucian Stanciu
Constantin Paleologu
Felix Albu
Keywords: Stereophonic Acoustic Echo Cancellation (SAEC); Recursive Least-Squares (RLS) algorithm; Data-Reuse (DR); Dichotomous Coordinate Descent (DCD); Widely Linear (WL) model.
Abstract:
The Stereophonic Acoustic Echo Cancellation (SAEC) setup implies the estimation of four loudspeaker-to-microphone unknown impulse responses, which generate unwanted acoustic replicas (i.e., echoes). In this context, some state-of-the-art approaches that combine the Widely Linear (WL) model with various versions of the Recursive Least-Squares (RLS) algorithm have been recently proposed. This paper focuses on the most recent one – the version that uses a Data-Reuse (DR) approach over the WL-RLS based on Dichotomous Coordinate Descent (DCD) iterations, i.e., the namely WL-DR-RLS-DCD. The target of the paper is to present how this algorithm behaves in various conditions, for input signals that present a very high correlation, such as speech or Auto-Regressive (AR) sequences. Simulations results proved that the DR approach is suitable in the SAEC context.
Pages: 6 to 9
Copyright: Copyright (c) IARIA, 2023
Publication date: April 24, 2023
Published in: conference
ISSN: 2308-3964
ISBN: 978-1-68558-034-6
Location: Venice, Italy
Dates: from April 24, 2023 to April 28, 2023