Analysis of Affine Projection Normalized Correlation Algorithm

Abstract

This paper proposes and develops a statistical analysis of Affine Projection Normalized Correlation Algorithm (AP-NCA) which is a combination of the Affine Projection Algorithm (APA) and Normalized Correlation Algorithm (NCA) for use in complex-domain adaptive filters. For impulse noise, two types are considered: one is found in observation noise and another at filter input. Through experiments with simulations and theoretical calculations of filter convergence, we demonstrate that the proposed AP-NCA is effective in making adaptive filters converge faster when the filter inputs are correlated, and robust in impulsive noise environments. It is observed that the theoretical convergence curves generally exhibit good agreement with the simulation results which shows that the analysis is valid for practical use.