Modeling of a Robust and Fast Noise Cancellation System

摘要

This paper presents a simple modified modeling of normalized least mean square (NLMS) algorithm with improved convergence speed and inherent robustness at the expense of less granularity for acoustic noise cancelling applications in Simulink environment. The paper further proposes that with an optimal choice of the crucial step size parameter, one can relatively guarantee faster convergence and conditions for robustness. This modification in the system can significantly be applied to any noise cancellation system. The designed system proved to be indubitably successful and functional in performing the intended application that is eliminating noise from a signal such as speech signal or any other type of audio signal. We stimulate the adaptive filter in MATLAB and analyze the performance of the algorithm in terms of convergence speed, computational complexity and stability. The simulation results are included to demonstrate and support the claims and points raised in the paper.