Variable tap-length non-parametric variable step-size NLMS adaptive filtering algorithm for acoustic echo cancellation

In teleconferencing and communication systems, the use of Loudspeaker Enclosure Microphone device leads to undesired echoes. To reduce these echoes, an acoustic echo canceller is used. In this paper, we propose a Variable Tap-length Non-Parametric Variable Step-Size NLMS (VT-NPVSS-NLMS) algorithm ba...

Full description

Saved in:
Bibliographic Details
Published in:Applied acoustics Vol. 159; p. 107074
Main Authors: Pauline, S. Hannah, Samiappan, Dhanalakshmi, Kumar, R., Anand, Ankita, Kar, Asutosh
Format: Journal Article
Language:English
Published: Elsevier Ltd 01.02.2020
Subjects:
Mean square error (MSE)
NLMS
Variable step-size
Variable tap-length
NLMS
Mean square error (MSE)
Variable tap-length
Variable step-size
ISSN:0003-682X, 1872-910X
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In teleconferencing and communication systems, the use of Loudspeaker Enclosure Microphone device leads to undesired echoes. To reduce these echoes, an acoustic echo canceller is used. In this paper, we propose a Variable Tap-length Non-Parametric Variable Step-Size NLMS (VT-NPVSS-NLMS) algorithm based on adaptive filtering for acoustic echo cancellation. The step-size is adjusted without the need for tuning too many parameters, using only the square of the average autocorrelation of a priori and a posteriori estimates of error. Moreover, the tap-length is varied to facilitate a high convergence speed and a small bias in tap-length from the optimum length. Hence, such a combination of the variable step-size algorithm with a variable tap-length provides faster convergence and reduced steady-state mean square error. The performance of the proposed algorithm is evaluated in terms of steady-state and transient mean square error. The advantages of the proposed algorithm, as compared to other adaptive algorithms, are presented using simulation results. From the results, we infer that for the VT-NPVSS-NLMS, the convergence speed is increased and the steady-state mean square error is reduced when compared with the conventional NLMS and the variable-step-size NLMS algorithm with a fixed tap-length.
ISSN:0003-682X
1872-910X
DOI:10.1016/j.apacoust.2019.107074