Channel Sparsity Aware Function Expansion Filters Using the RLS Algorithm for Nonlinear Acoustic Echo Cancellation

Jean Jiang,Vinith Vijayarajan,Lizhe Tan

doi:10.1109/access.2020.3005408

Jean Jiang, Vinith Vijayarajan + Show 1 more

Open Access

https://doi.org/10.1109/access.2020.3005408

Copy DOI

Abstract

In this paper, we propose a channel sparsity aware sequential recursive least squares (sparse SEQ-RLS) algorithm for function expansion filters with applications in nonlinear echo cancellation. The algorithm is developed based on a diagonal channel structure from the Volterra filter and updating dominant coefficients taking into consideration of sparse elements in the diagonal channel. The third-order Volterra, third-order even mirror Fourier nonlinear (EMFN), and functional link artificial neural network (FLANN) filters are developed according to the sparse SEQ-RLS algorithm. The computation complexity for the upper bound is analyzed to validate the efficiency for each proposed filter. Computer simulation results demonstrate that all proposed function expansion filters with the sparse SEQ-RLS algorithm are effective for nonlinear echo cancellation. In general, the EMFN filter provides better performance compared to the Volterra and FLANN filters.

Highlights

Speech quality is in demand for voice commanded systems and telephony [1]–[4]
We develop function expansion adaptive filters wielding a new channel sparsity-aware recursive least squares (RLS) algorithm using a sequential update
We have developed the function expansion adaptive filters by applying the sparse sequential RLS (SEQ-RLS) algorithm for nonlinear acoustic echo cancellation

Summary

INTRODUCTION

Speech quality is in demand for voice commanded systems and telephony [1]–[4]. The voice communication system in real time often suffers from audible echoes. J. Jiang et al.: Channel Sparsity Aware Function Expansion Filters Using the RLS Algorithm the nonlinear echo path are not significant and their values are close to zero. The developed nonlinear adaptive filters using a sparse sequential RLS (SEQ-RLS) algorithm adopt a discard function to neglect the coefficients whose values are close to zero in the weight vector for each filter channel in order to reduce the computational load and to improve the algorithm convergence rate. FUNCTION EXPANSION FILTERS In order to model a nonlinear echo path due to signal companding and/or due to over driven amplifiers to near saturation, we apply the sparse SEQ-RLS algorithm to the following functional expansion filters. To validate the developed algorithms, we first perform nonlinear system identification and compare the performances with the Volterra, FLANN and EMFN adaptive filters each using the sparse SEQ-RLS algorithm respectively. Environment condition, the discard function is not preferred unless further reduction of computational load is a must

SINGLE TALK SIMULATIONS

CONCLUSION