분산 추정을 위한 아크탄젠트 기반의 Diffusion NLMS 알고리즘

Abstract

Popular diffusion adaptive filter algorithms based on mean-square can not be robust in impulsive noise environments. To overcome such difficulty, a arctangent-cost-function based diffusion NLMS is proposed in this paper. The proposed algorithm can be approximated by the NLMS algorithm in case of no impulsive noise or by zero update when the impulsive interference occurs. Finally, it is demonstrated through system identification simulations that the proposed algorithm yields better MDS performance rather than other diffusion based adaptive filtering algorithms.