A new strategy for auxiliary noise injection in narrowband active noise control

Abstract

Injecting auxiliary noise is an effective way of implementing online secondary-path modeling in active noise control (ANC) systems. The auxiliary noise may be scaled or multiplied by absolute value of one-sample-delayed residual noise signal. This strategy works well but more effort is needed to further reduce contribution of the injected auxiliary noise to the residual noise power. In this paper, we propose a novel strategy for the auxiliary noise injection in a typical narrowband ANC system, which not only significantly improves system convergence but also considerably reduce the steady-state residual noise power. The auxiliary noise to be injected is scaled by a signal which is obtained by passing a nonlinear function of one-sample-delayed residual noise signal through a lowpass filter. In the early stage of adaptation a large scaled auxiliary noise is injected to excite the secondary path to favor the secondary-path modeling, while in the steady state the scaled auxiliary noise becomes considerably small and hence contributes significantly less to the residual noise power. Extensive simulations demonstrate the effectiveness and superiority of the proposed strategy.