System Identification Using the Second Order MLMS Algorithm

제2차 MLMS 알고리즘을 이용한 시스템 Identification

This paper analyzes the properties of such algorithm that corresponds to the LMS algorithm with additional update terms, parameterized by the scalar factors $\alpha$ and $\beta$, and presents its structure. The analysis of convergence leads to complex eigenvalues of the transition matrix for the mean weight vector. Regions in which the algorithm becomes stable are demonstrated. The computational cmomplexities of MLMS algorithms are compared with those of MADF, sign and the conventional LMS algorithms. In application of the system identification the second order momentum MLMS algorithm has faster convergence speed than LMS and the first order MLMS algorithms.