Robust variable step-size sign subband adaptive filter algorithm against impulsive noise

Abstract

The main drawback of the sign subband adaptive filter algorithm (SSAF) is its large steady-state error and poor tracking capability. A robust variable step-size SSAF algorithm is presented to improve the performance, which uses the optimal step-size for each subband by minimizing the mean square deviation. Since the step-size contains the variance of noise-free priori error, the shrinkage denoising method is used to estimate the noise-free error signal. With the step-size adaptation influenced by the variance of error signal for each subband, the proposed algorithm is shown to achieve a significant improvement in both the convergence rate and the steady-state error than the SSAF algorithms. Besides, the proposed algorithm offers robust performance with respect to impulsive noise and good ability of tracking unknown system. Finally, simulation results demonstrate that these features of the proposed algorithm in system identification and echo-cancellation applications.