An intelligent momentum perturbed variable step-size adaptive algorithm for fast converging HNANC systems

• Published in: Digital signal processing 2024-02, Vol.145, p.104305, Article 104305
• Main Author: Kar, Asutosh
• Format: Article
• Language: English
• Subjects: Adaptive filter; Hybrid active noise cancellation; Least mean square; Mean noise reduction; Mean square error; Performance analysis
• ISSN: 1051-2004; 1095-4333
• DOI: 10.1016/j.dsp.2023.104305

The existing methodologies for hybrid narrow-band active noise control (HNANC) systems utilises variants of the FXLMS algorithm, which typically resulted in degraded convergence and steady-state performance. In this paper, a class of variable step size (VSS) adaptive algorithms are explicitly developed for a HNANC system analyses and the convergence as well as the steady-state performance of the proposed framework are evaluated in comparison to the existing state-of-the-art. For an HNANC system designed to operate in a multiple noise environment, the VSS filtered-x LMS (VSSFXLMS) and momentum VSS filtered- x weight accumulated least mean square (MVSS-FXWALMS) algorithms are proposed. The proposed algorithm employs a variable step size that is inspired by the variable momentum factor, and combines these two parameters to drive the FXWALMS algorithm for an HNANC system in order to accomplish maximum noise reduction. In a multiple noise environment setup for an HNANC system, the proposed algorithm surpasses the state-of-the-art in nearly all conventional operating conditions. A detailed performance analysis of the proposed algorithm has been carried out using mean noise reduction ratio, steady-state power spectra at different conditions. The results demonstrate a significant improvement in noise reduction performance compared to the counterparts.