SSP'05 -- Proceedings

	IEEE/SP 13th workshop on Statistical Signal Processing July, 17-20, 2005 - Bordeaux - France

Information regarding the paper

Title

Geometric analysis of filtered-x LMS algorithms

Author(s)

Dayong Zhou	University of Oklahoma
Victor DeBrunner	University of Oklahoma
Linda DeBrunner	University of Oklahoma
Yunhua Wang	University of Oklahoma

Get the paper in PDF format

To obtain Acrobat Reader (version 5 minimum required) necessary to his read.

Abstract

In this paper, we present a geometric based analysis of the “filtered” LMS algorithms such as the filtered-x LMS algorithm and the adjoint LMS algorithm. Using this method, the maximum step size, the effects of the secondary path estimation error and the relationship between the adjoint, secondary path equalization and FXLMS LMS algorithms can be easily obtained. Furthermore, our introduced method yields these conditions with some geometrical meaning and intuitive explanation. Based on our introduced method, we also discuss the convergence speed of the filtered-x LMS algorithm for the first time.