Maximum total generalized correntropy adaptive filtering for parameter estimation

Abstract

In this study, we consider the parameter estimation problem for an errors-in-variables (EIV) model with impulse noise. New adaptive filtering, called the maximum total generalized correntropy (MTGC) adaptive filtering algorithm, is developed to further improve the robustness of conventional adaptive filtering algorithms. Specifically, the proposed approach is derived by integrating the generalized maximum correntropy (GMC) criterion into the total least square (TLS) framework. The local stability and steady-state properties are investigated with the aid of the generalized Gaussian process. Numerical simulations are presented to illustrate the effectiveness of the proposed method in the presence of impulse noise.