A Modified Filtered-x LMAT Algorithm for Active Noise Control of Impulsive Noise

In this era, the reduction of impulsive noise within the active noise control (ANC) has become a serious problem. In literature, a number of techniques has been offered for ANC systems. Impulsive noises have heavy tails and they are modelled using the non-Gaussian stable process for which variance doesn't exist. Filtered-x least mean square (FxLMS) algorithm is a well celebrated algorithm but it becomes unsteady in the presence of large amplitude impulses. Adaptive filters established on the higher order error power (HOEP) principle possess enhanced filtering capability than least mean square (LMS) family algorithms for non-Gaussian noises. The performance of least mean absolute third (LMAT) algorithm, which reduces the cube power of absolute error, is tested under ANC system and thus, filtered-x least mean absolute third algorithm (FxLMAT) is proposed. To further refine the convergence and stability of FxLMAT algorithm, two modifications namely input samples ignored FxLMAT (SIFxLMAT) and input samples clipped FxLMAT (SCFxLMAT) algorithm are proposed, which modify the error and reference signals on the basis of statistical parameters of input noise. Simulation outcomes illustrate that the proposed SCFxLMAT algorithm possesses better stability and improved convergence than that of FxLMAT and SIFxLMAT algorithm.