Incorporating Nonparametric Knowledge to the Least Mean Square Adaptive Filter

Robust diffusion LMS over adaptive networks

“…Probabilistic DLMS proposed in [36] approximates the posterior distribution with an isotropic Gaussian distribution. Recently, the non-parametric probabilistic least mean square (NPLMS) adaptive filter has been proposed in [37] for the estimation of an unknown parameter vector from noisy measurements. The NPLMS combines parameter space and signal space by combining the prior knowledge of the probability distribution of the process with the evidence existing in the signal.…”

“…Utilizing pseudo-Huber loss function in Diffusion LMS has been investigated in [39] to create a robust algorithm against noise in adaptive networks. Therefore, the main contribution of this manuscript is to extend the NPLMS algorithms [37] over distributed adaptive networks and benefit pseudo-Huber loss function [39] to design the likelihood function.…”