A Parallel Kernelized Data-Reusing Maximum Correntropy Algorithm

Abstract

In this brief, a parallel kernel data-reusing maximum correntropy (PKDRMC) algorithm is proposed within the framework of nonlinear adaptive filtering. The PKDRMC algorithm consists of two branches, namely, the data-reusing maximum correntropy algorithm, and its kernelized form to combat the non-Gaussian interferences. Then, a new cost function is put forward based on the two different schemes, and it is investigated via nonlinear channel equalization (NCE). The proposed PKDRMC algorithm is robust against impulsive-noise environments. Simulations in the NCE under impulsive-noise settings perform that the PKDRMC algorithm outperforms the popular kernel adaptive algorithms.