Noise cancellation using nonlinear adaptive FIR digital filters

Abstract

The authors present the structure and the adaptive algorithm of a novel nonlinear adaptive FIR (finite impulse response) digital filter for cancelling broadband noise from the signal while preserving the sharp edges of the signal. The filter structure is constructed by decomposing a FIR digital filter into N parallel subfilters, and nonlinear sigmoid functions are placed at the end of each subfilter as well as after the summation of all subfilter outputs. Two levels of adjustable nonlinear functions introduced play a significant role in cancelling broadband noise from signals with sharp edges. Simulation results show that the novel nonlinear adaptive FIR digital filter outperforms the linear adaptive FIR digital filter and the nonlinear median filter in broadband noise cancellation of signals. >