Low Complexity Implementation Method for the Adaptive Filters based on the Gaussian Model

This paper proposes a method to lower the computational complexity of the adaptive filters based on the Gaussian model (GM-ADF). The conventional GM-ADF is shown to be robust against impulsive noise. However, the price to pay is a higher computational complexity since each coefficient in the GM-ADF has its own Gaussian model, and the parameters (means and variances) of the Gaussian distributions must be computed to estimate them. In this paper, we propose a method which trims down the number of coefficients modeled by the Gaussian distributions, so that the number of the parameter calculated per iteration reduces. Besides, we alter the way to update the coefficients, which would possibly increase the rate of convergence under some conditions. The performance of the proposed method is demonstrated by computer simulations.