Novel active noise control based on a robust filtered-x normalized least mean square sign algorithm against large measurement and impulsive noises

Abstract

This paper presents a novel active noise control (ANC) based on a robust filtered-x normalized least mean square sign (R-FxNLMSS) algorithm against the large measurement noises and impulsive noises. The R-FxNLMSS algorithm updates the filter using the Euclidean norm of the sum from the previous weight vectors to the present weight vectors, which has robustness not only against the large measurement noises but also against the impulsive noises. Simulation results show that the proposed ANC based on the R-FxNLMSS algorithm has lower steady-state errors and faster convergence rate than the ANC based on the existing algorithms in extreme environments where the measurement noises are very large and the impulsive noises are generated randomly. © 2020 Institute of Control, Robotics, and Systems - ICROS.