On family of fractional lower order moment (FLOM)-based algorithms for active noise control of impulsive noise sources

Abstract

Active noise control (ANC) is based on the principle of destructive interference of propagating acoustic waves; essentially a cancelling signal is generated and combined with the primary noise to achieve acoustic cancellation around location of the error microphone. In this paper, we consider a very challenging application of ANC for impulsive noise. The impulsive noise can be modeled using non-Gaussian stable process for which second order moment does not exist. The most famous filtered-x-LMS (FxLMS) algorithm for ANC systems, is based on minimization of the variance of the error signal, and therefore, becomes unstable for the impulsive noise. It has been shown that filtered-x least mean p-power (FxLMP) algorithm; based on minimizing the fractional lower order moment (FLOM) that does exist for stable distributions; gives robust performance for impulsive ANC. However, the convergence speed of the FxLMP algorithm is very slow. Recently; we have proposed various variants of FxLMP algorithm, so that an improved convergence and noise reduction performance is achieved. In this paper, we propose modifying and employing generalized normalized LMP algorithm (GNLMP) algorithm for ANC of impulsive noise. The computational complexity of proposed algorithm is comparable to the existing FLOM-based ANC algorithms. Extensive simulations are carried out, which demonstrate the effectiveness of proposed algorithm. We observe that, in comparison with the existing FLOM-based ANC algorithms, the proposed algorithm gives best performance for ANC of impulsive noise sources.