An Active Noise Control Algorithm Using Diffusion Adaptive Filter Robust to Impulsive Noise and Input Noise in Sparse Systems*

This paper introduces an advanced active noise control (ANC) algorithm formulated within a diffusion normalized least mean square (D-NLMS) framework, designed to be highly resilient to both impulsive disturbances and input noise in sparse environments. To address secondary path distortion issues commonly encountered in ANC systems, the algorithm employs a Filtered-x configuration. Robustness to impulsive noise is further improved through the adoption of a dynamic step-size scaling mechanism and a modified Huber error function. In addition, a bias compensation strategy is analytically derived to effectively counteract the steady-state bias caused by input noise. By incorporating an L0-norm regularization term into its cost function, the proposed approach enhances adaptability to sparse system characteristics. Simulation results validate that the proposed ANC method offers superior steady-state accuracy, faster convergence, and enhanced robustness compared to traditional techniques when facing impulsive and input noise.