Analysis of the frequency-domain block LMS algorithm

Author

Chan, Kheong Sann ; Farhang-Boroujeny, B.

Author_Institution

Dept. of Electr. Eng., Nat. Univ. of Singapore, Singapore

Volume

1

fYear

2000

fDate

2000

Firstpage

380

Abstract

We present a new rigorous analysis of the FBLMS algorithm. We derive the equations which control the expected convergence of the FBLMS algorithm and then go on to analyze the eigenvalues of the matrix which control the convergence behaviour of the FBLMS algorithm. Using Gerschgorin´s theorem, we then show that all the eigenvalues of the normalized constrained and unconstrained FBLMS algorithms tend asymptotically towards the same value

Keywords

adaptive signal processing; convergence of numerical methods; eigenvalues and eigenfunctions; frequency-domain analysis; least mean squares methods; matrix algebra; FFT; Gerschgorin´s theorem; adaptive filters; algorithm convergence; equations; frequency-domain block LMS algorithm; matrix eigenvalues; normalized constrained FBLMS algorithm; unconstrained FBLMS algorithm; Adaptive filters; Algorithm design and analysis; Constraint theory; Convergence; Costs; Eigenvalues and eigenfunctions; Equations; Frequency domain analysis; Least squares approximation; Time domain analysis;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 2000. ICASSP '00. Proceedings. 2000 IEEE International Conference on

Conference_Location

Istanbul

ISSN

1520-6149

Print_ISBN

0-7803-6293-4

Type

conf

DOI

10.1109/ICASSP.2000.861982

Filename

861982