A strict stability limit for adaptive gradient type algorithms

Authors:
Robert Dallinger;Markus Rupp
Affiliations:
Institute of Communications and Radio-Frequency Engineering, Vienna University of Technology, Vienna, Austria;Institute of Communications and Radio-Frequency Engineering, Vienna University of Technology, Vienna, Austria
Venue:
Asilomar'09 Proceedings of the 43rd Asilomar conference on Signals, systems and computers
Year:
2009

Citing 3
Cited 0

L2-Gain and Passivity in Nonlinear Control

L2-Gain and Passivity in Nonlinear Control
A time-domain feedback analysis of filtered-error adaptive gradientalgorithms

IEEE Transactions on Signal Processing
Error-energy bounds for adaptive gradient algorithms

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper considers gradient type algorithms with a regression vector allowed to be different from the input vector. To cover the most general case, no restrictions are imposed on the dependency between the excitation vector and the regression vector. In terms of l2--stability, for the real-valued domain, a convergence analysis is performed based on the singular value decomposition. It reveals that such algorithms are potentially unstable if the input vector and the regression vector do not have the same direction. For the conventional gradient type algorithm (for which latter vectors are parallel), an l2--stability bound, known from literature to be sufficient, can be shown to be actually strict. Simulations demonstrate how the presented method can be used to discover unstable modes of an apparently stable algorithm.