Tracking Performance of Improved Convex Combination Adaptive Filter Based on Maximum Correntropy Criterion

A convex combination adaptive filter based on maximum correntropy criterion (CMCC) was widely used to solve the contradiction between the step size and the misadjustment in impulsive interference. However, one of the major drawbacks of the CMCC is its poor tracking ability. In order to solve this problem, this paper proposes an improved convex combination based on the maximum correntropy criterion (ICMCC), and investigates its estimation performance for system identification in the presence of non-Gaussian noise. The proposed ICMCC algorithm implements the combination of arbitrary number of maximum correntropy criterion (MCC) based adaptive filters with different adaption steps. Each MCC filter in the ICMCC is capable of tracking a specific change speed, such that the combined filter can track a variety of the change speed of weight vectors. In terms of normalized mean square deviation (NMSD) and tracking speed, the proposed algorithm shows good performance in the system identification for four non-Gaussian noise scenarios.