Discrete-Time Dithering Technique for Designing Time-Varying Signal Adapted Filter Banks

Author

Vouras, Peter G. ; Tran, Trac D.

Author_Institution

Johns Hopkins Univ., Baltimore

fYear

2007

fDate

14-16 March 2007

Firstpage

838

Lastpage

843

Abstract

This paper describes the use of sinusoidal dithers in a lattice filter bank structure to create a time-varying paraunitary filter bank that approximates and tracks in near real-time the ideal principal component filter bank corresponding to a nonstationary random input process. The sinusoidal dithers are used to estimate the gradient of a cost function which is minimized using non-linear minimization techniques.

Keywords

discrete time systems; lattice filters; minimisation; time-varying filters; cost function estimation; discrete-time dithering; lattice filter bank structure; nonlinear minimization; nonstationary random input process; principal component filter bank; sinusoidal dithers; time-varying paraunitary filter bank; time-varying signal adapted filter bank design; Channel bank filters; Cost function; Eigenvalues and eigenfunctions; Filter bank; Finite impulse response filter; Karhunen-Loeve transforms; Lattices; Matrix decomposition; Signal design; Signal synthesis; dithers; paraunitary filter banks; principal component filter banks;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Sciences and Systems, 2007. CISS '07. 41st Annual Conference on

Conference_Location

Baltimore, MD

Print_ISBN

1-4244-1063-3

Electronic_ISBN

1-4244-1037-1

Type

conf

DOI

10.1109/CISS.2007.4298425

Filename

4298425