Basis Expansion Adaptive Filters for Time-Varying System Identification

Author

Rugini, Luca ; Leus, Geert

Author_Institution

Fac. of Electr. Eng., Delft Univ. of Technol., Delft

fYear

2007

fDate

12-14 Dec. 2007

Firstpage

153

Lastpage

156

Abstract

In this paper, we extend the concept of block adaptive filters to what we call basis expansion adaptive filters. While in block adaptive filters the system is assumed to be constant within a block, our basis expansion adaptive filters model the time variation of the system within a block by a set of basis functions. This allows us to improve the tracking performance of block adaptive filters considerably. We focus on stochastic gradient type of adaptive filters, although extensions to other types of adaptive filters can be envisioned.

Keywords

adaptive filters; filtering theory; gradient methods; identification; least mean squares methods; linear systems; time-varying systems; basis expansion adaptive filters; basis function; block adaptive filter; linear time-varying system identification; stochastic gradient; Adaptive algorithm; Adaptive filters; Convergence; Filtering algorithms; Least squares approximation; Polynomials; Resonance light scattering; Stochastic systems; System identification; Time varying systems; Basis expansion model; block adaptive filters; linear time-varying systems;

fLanguage

English

Publisher

ieee

Conference_Titel

Computational Advances in Multi-Sensor Adaptive Processing, 2007. CAMPSAP 2007. 2nd IEEE International Workshop on

Conference_Location

St. Thomas, VI

Print_ISBN

978-1-4244-1713-1

Electronic_ISBN

978-1-4244-1714-8

Type

conf

DOI

10.1109/CAMSAP.2007.4497988

Filename

4497988