Robust feedback active noise control algorithm for impulsive additive noise

Abstract

Active noise control (ANC) is a method to reduce the unwanted noise level by introducing secondary noise which has the same amplitude and antiphase with the primary noise. The feedback ANC algorithm is a kind of linear predictor and it uses only one sensor to get the information about the error signal and reference signal. The filtered-x LMS (least mean square) algorithm is widely used for implementing the ANC system because of its simplicity and good performance. But when the additive noise is impulsive, the performance of LMS-type algorithms are known to be lowered. In this paper, the robust feedback ANC algorithm is proposed when impulsive additive noise is present. Instead of the LMS algorithm, the proportion-sign algorithm (PSA) which is a mixture of LMS and dual sign algorithm (DSA) is used to estimate the primary noise field robustly when impulsive additive noise is present. Computer simulations were performed under impulsive additive noise circumstances and showed better performance with the proposed algorithm than that with the conventional LMS algorithm.