Passband disturbance reduction in periodic active noise control systems

For narrowband active noise control (ANC), the filtered-X least-mean-square (LMS) algorithm with a sinusoidal reference signal forms an adaptive notch filter. Unfortunately, the effect of the secondary path on the uncorrelated noise is to introduce an undesired passband disturbance. To reduce this disturbance problem, this paper analyzes the passband amplification condition and develops an algorithm that consists of two interconnected adaptive filters using the same internally generated reference signal. The first adaptive filter eliminates the uncorrelated noise components in the residual error, and its output is used to update the second adaptive filter for noise control. Computer simulations were conducted to verify the analysis results and to demonstrate the disturbance reduction for periodic ANC while using a relatively large step size to improve the convergence rate.