The MMFxLMS algorithm for active noise control with on-line secondary path modelling

Abstract

This paper presents an architecture namely the Mirror MFx (MMFx) for adapting adaptive filtering algorithms for Active Noise Control (ANC) with on-line secondary path modelling. The proposed architecture is used in conjunction with the LMS algorithm, resulting in the MMFxLMS algorithm. A time domain analysis of the algorithm is presented, showing that the algorithm converges regardless of secondary path modeling errors. Simulations of the algorithm in different conditions but with the same parameters result in 100% convergence. The algorithm is especially suited to deal with large and sudden changes in the secondary path when the ANC system is in operation. Comparisons with competing algorithms are made, showing that they do not reach the same performance.