Noncircular Complex ICA by Generalized Householder Reflections

Author

Xi-Lin Li ; Adali, Tulay

Author_Institution

Fortemedia Inc., Sunnyvale, CA, USA

Volume

61

Issue

24

fYear

2013

fDate

Dec.15, 2013

Firstpage

6423

Lastpage

6430

Abstract

We develop an efficient algorithm for noncircular complex independent component analysis (ICA) by optimizing the rows of separation matrix independently using a generalized Householder reflector and Newton method. We apply the new algorithm to the separation of complex generalized Gaussian distributed (GGD) sources, and propose a convenient way to learn the noncircularity of the sources. Simulation results are reported to study the convergence behavior, and confirm the superior performance of the new algorithm.

Keywords

Gaussian processes; blind source separation; independent component analysis; matrix algebra; Newton method; blind source separation; complex generalized Gaussian distributed sources; generalized householder reflections; noncircular complex ICA; noncircular complex independent component analysis; separation matrix; Algorithm design and analysis; Independent component analysis; Jacobian matrices; Optimization; Signal processing algorithms; Transforms; Vectors; Blind source separation; Newton algorithm; Wirtinger calculus; generalized Gaussian distribution (GGD); independent component analysis; noncircularity;

fLanguage

English

Journal_Title

Signal Processing, IEEE Transactions on

Publisher

ieee

ISSN

1053-587X

Type

jour

DOI

10.1109/TSP.2013.2286113

Filename

6636089