Title :
Joint diagonalization: Is non-orthogonal always preferable to orthogonal?
Author :
Souloumiac, Antoine
Author_Institution :
Stochastic Processes & Spectra Lab., CEA Saclay, Gif-sur-Yvette, France
Abstract :
This paper deals with the efficiency of orthogonal versus non-orthogonal joint diagonalization techniques for blind source separation. The question is: are non-orthogonal algorithms always preferable to orthogonal ones in practical noisy conditions? The two techniques are compared in terms of separation performance through numerical simulations with different levels of additive noise and numbers of samples. Both fourth-order and second-order blind source separation contexts are investigated.
Keywords :
blind source separation; independent component analysis; additive noise; fourth-order blind source separation; independent component analysis; nonorthogonal joint diagonalization techniques; numerical simulations; orthogonal joint diagonalization techniques; second-order blind source separation; Additive noise; Blind source separation; Conferences; Covariance matrix; Decorrelation; Laboratories; Sensor phenomena and characterization; Signal processing; Source separation; Stochastic processes; J-Di; JADE; Non-orthogonal joint diagonalization; SOBI; blind source separation; independent component analysis; orthogonal joint diagonalisation; pre-whitening;
Conference_Titel :
Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP), 2009 3rd IEEE International Workshop on
Conference_Location :
Aruba, Dutch Antilles
Print_ISBN :
978-1-4244-5179-1
Electronic_ISBN :
978-1-4244-5180-7
DOI :
10.1109/CAMSAP.2009.5413273
