A study of active impulsive noise control based on an adjustable fractional cost function

Area of Science:

• Acoustics
• Signal Processing
• Control Systems

Background:

• Classic filtered-x least mean square (FxLMS) algorithms struggle with impulsive noise.
• Existing algorithms are primarily designed for instantly decaying impulses, not real-world impulsive noise with finite decay times.
• The performance of current methods on impulsive noise with finite decay times is largely unverified.

Purpose of the Study:

• To propose an enhanced FxLMS algorithm for active noise control (ANC) of impulsive noise.
• To address the challenges posed by the non-Gaussian distribution of impulsive noise in ANC systems.
• To improve the convergence speed and adaptability of ANC systems for various noise types.

Main Methods:

• Replaced the standard cost function with an adjustable fractional function for nonlinear error signal compression.
• Introduced an adjustable compression factor to tailor the error function to different impulsive noise intensities.
• Implemented a time-varying normalized function for adaptive step-size adjustment to accelerate filter iteration convergence.

Main Results:

• The proposed algorithm demonstrates superior performance in controlling impulsive noise with finite decay times compared to existing ANC algorithms.
• It also outperforms current methods in suppressing instantly decayed impulsive noise.
• The enhanced algorithm shows improved effectiveness against broadband Gaussian noise.

Conclusions:

• The enhanced FxLMS algorithm offers a robust solution for active noise control of impulsive noise, particularly those with finite decay characteristics.
• The nonlinear compression and adaptive step-size adjustments contribute to better performance and faster convergence.
• This approach provides a significant improvement over existing ANC techniques for a wider range of noise conditions.