Title :
A parallel weight extraction method for sliding window block RLS
Author :
Sakai, Hideaki ; Kuroda, Manabu ; Ohno, Shuichi
Author_Institution :
Kyoto Univ., Japan
Abstract :
The square-root-free algorithm for sliding window block RLS (recursive least squares) based on the up/down-dating Cholesky factorization via vectorized Gram-Schmidt pseudo-orthogonalization of S.-F. Hsieh et al. (1991) is used to derive an algorithm for extracting the optimal weights of this RLS. Its implementation by a rhombic systolic array is shown. Unlike the arrays for RLS with a forgetting factor, there are only two kinds of cells in the array considered.<>
Keywords :
adaptive filters; least squares approximations; parallel algorithms; systolic arrays; Cholesky factorization; cells; optimal weights; parallel weight extraction; rhombic systolic array; sliding window block RLS; square-root-free algorithm; vectorized Gram-Schmidt pseudo-orthogonalization;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1993. ICASSP-93., 1993 IEEE International Conference on
Conference_Location :
Minneapolis, MN, USA
Print_ISBN :
0-7803-7402-9
DOI :
10.1109/ICASSP.1993.319521
