SSP'05 -- Proceedings

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

Information regarding the paper

Title

An Improved Recursive Least Squares Algorithm Robust to Input Power Variation

Author(s)

Charles S. Ludovico	State University of Londrina
José Carlos Bermudez	Federal University of Santa Catarina

Get the paper in PDF format

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

Abstract

This paper proposes a new recursive least-squares adaptive algorithm that improves the steady-state performance of the recently proposed Variable Memory Length (VML) algorithm. Most RLS-type algorithms tend to increase the error in the estimated weight vector during reduced power situations. Like VML, the new algorithm, called Robust VML (RVML), is robust in system identification applications in which the input power is significantly reduced during operation. The RVML algorithm, however, improves the robustness of the VML algorithm when there is significant input power variations during convergence. It should encounter application in systems such as automotive suspension fault detection and adaptive control, and system identification using speech signals. In both cases, considerable periods of power variation during operation are common.