Recursive MCC Algorithm with Proportionate Update Mechanism for Robust Sparse System Identification

Abstract

This paper proposed an effective robust adaptive filtering algorithm (RAFA) for sparse system identification (SSI) considering impulsive noise interference. Because of the robustness of the correntropy, it has been utilized as a cost, called maximum correntropy criterion (MCC), for designing robust sparse AFA via proportionate update method. However, the original proportionate MCC algorithm has some defects in the initial convergence speed. To address this problem, the proportionate update matrix is inserted into the gain matrix of the recursive MCC algorithm, which takes advantage of both proportionate update method and recursive RAF =A. Therefore, the proposed proportionate recursive MCC (PRMCC) algorithm not only can obtian low steady-state misalignment, but also can show a fast convergence speed for SSI in presence of impulsive noise. Some numerical simulations are performed to test the robustness of the proposed PRMCC method under different conditions, and the simulation results show that it can outperform other outstanding algorithms for SSI in non-Gaussian noise with outliers environments.