Title :
A new normalized signed regressor LMS algorithm
Author :
Takahashi, Kiyoshi ; Mori, Shinsaku
Author_Institution :
Keio Univ., Yokohama, Japan
Abstract :
The normalized signed regressor algorithm is the NLMS algorithm based on clipping of the input samples which are elements of the input data vector. In the new algorithm, a clipped sample is used to update coefficients when the absolute value of the sample is larger than the average of the absolute values of the input samples. Analysis shows that the proposed algorithm has better convergence characteristics than the conventional normalized signed regressor algorithm
Keywords :
adaptive filters; convergence of numerical methods; digital filters; filtering and prediction theory; least squares approximations; clipping; convergence characteristics; normalized signed regressor LMS algorithm; Adaptive filters; Adaptive systems; Algorithm design and analysis; Convergence; Eigenvalues and eigenfunctions; Estimation error; Finite impulse response filter; Least squares approximation; Magnetooptic recording; System identification;
Conference_Titel :
Singapore ICCS/ISITA '92. 'Communications on the Move'
Print_ISBN :
0-7803-0803-4
DOI :
10.1109/ICCS.1992.255075