A New Diffusion Filtered-X Affine Projection Algorithm: Performance Analysis and Application in Windy Environment

Wind noise caused by turbulent flows over microphones usually has detrimental impacts on the reference signal of an active noise control (ANC) system and degrade its performance considerably. This paper evaluates the influence of wind noise on the filtered-x affine projection algorithm (FxAPA) for ANC systems. To improve the performance of noise control systems in windy environments, the FxAPA is modified based on a wind-noise-free least mean squares error and the diffusion control technology that could suppress interruptions in the collected primary noise. Performance analysis of the resulting de- regularized diffusion FxAPA (Diff-FxAPA) algorithm has been carried out at the presence of wind noise, which reveals the underlying mechanism of the new algorithm on suppressing the wind noise interruption. Difference equations describing the mean and mean squares convergence behaviors of this ANC system are derived to characterize its optimal solution, estimation bias and variance, and convergence conditions. Computer simulations with real-recorded data validate the theoretical analysis and show that the noise reduction performance of conventional ANC methods degrades with wind noise while the proposed algorithm has a stable performance under a variety of windy conditions.