Adaptive Kernelized Subfilter Nonlinear AEC Algorithm

Abstract

The acoustic echo canceller developed based on a linear echo path suffers performance degradation due to distortion induced by the electronic components in hands-free devices. Recently kernel-based approach is proposed to minimize the effects of distortion. Here, the distortion is modeled using the kernel method. However, there exists a scope to further enhance the performance of echo cancellers. In this work, we propose an improved variant of the kernel approach for enhancing the performance by maintaining a better trade-off between the rate of convergence and steady-state. Simulation results show that the proposed approach was able to perform better than the existing kernel method in minimizing the impact of the sigmoidal-based saturation distortion on the echo cancellation performance.