Minimization of waterbed effect using a frequency-domain algorithm

In feedback active noise control, noise is typically amplified outside the desired frequency band. This phenomenon is called the waterbed effect and it is dependent on the sensitivity function of the system including the control filter and the secondary path. Using frequency-domain adaptation on the control filter, the undesired amplification can be reduced by introducing a constraint on the update algorithm. This paper concerns the minimization of the waterbed effect using a constrained frequency-domain algorithm. The maximum value of the magnitude of sensitivity function is constrained under a given limit. The constraint is combined with the frequency-domain FXLMS algorithm to obtain the constrained frequency-domain FXLMS algorithm. The performance of the resulting algorithm is simulated and analyzed. The simulation results indicate that the constraint reduces the amplification due to the waterbed effect.