On Active Impulsive Noise Control (AINC) Systems: Developing a Filtered-Reference Adaptive Algorithm Using a Convex-Combined Normalized Step-Size Approach

Research output: Contribution to journalArticlepeer-review

13 Citations (Scopus)

Abstract

This paper develops an efficient adaptive filtering algorithm for active impulsive noise control (AINC) systems. For AINC systems, the filtered-x least mean square (FxLMS) algorithm fails to converge due to the impulsive nature of the noise source. In previous work, the step-size of the FxLMS algorithm was normalized using the power estimate of the error as well as the reference signals, resulting in the improved normalized step-size FxLMS (INSS-FxLMS) algorithm. The INSS-FxLMS algorithm exhibits a robust performance for AINC systems; however, it uses a preselected fixed step-size. Therefore, the INSS-FxLMS algorithm results in a compromise between convergence speed and noise reduction. The proposed algorithm employs a convex-combined step-size (CCSS) within the framework of the INSS-FxLMS algorithm. While normalization takes care of the impulsive nature of noise, the CCSS solves the above-mentioned trade-off issue. Essentially, the CCSS selects a large (small) value of the step-size in the transient (steady) state of the AINC system. It is demonstrated by extensive computer simulations that the proposed algorithm outperforms the existing counterparts for a variety of case studies in AINC systems.

Original languageEnglish
Pages (from-to)4354-4377
Number of pages24
JournalCircuits, Systems, and Signal Processing
Volume39
Issue number9
DOIs
Publication statusPublished - Sept 1 2020

Keywords

  • Active impulsive noise control
  • Convex combination
  • Filtered-reference adaptive filtering
  • Stable distribution
  • Variable normalized step-size

ASJC Scopus subject areas

  • Signal Processing
  • Applied Mathematics

Fingerprint

Dive into the research topics of 'On Active Impulsive Noise Control (AINC) Systems: Developing a Filtered-Reference Adaptive Algorithm Using a Convex-Combined Normalized Step-Size Approach'. Together they form a unique fingerprint.

Cite this