How blind are we in Blind Signal Processing?

Conference contribution


Publication Details

Author(s): Kellermann W
Publication year: 2006
Pages range: 3046
Language: English


Abstract

Blind signal processing algorithms aim at blindly solving classical signal processing problems such as system identification, signal separation, or source localization. Here it is investigated to which extent some of the more popular blind signal processing concepts are really blind when they are used for signal acquisition in the acoustic domain. Blindness regarding the source configuration and the source signal properties, blindness with respect to channel properties, and, finally, blindness regarding the microphone configuration are investigated. It turns out that many algorithms are explicitly nonblind with respect to important properties of the source configuration, such as the number of sources and their pointlike nature. The source signals typically have to fulfill certain statistical properties or have to meet sparseness constraints. Some algorithms rely on a dominant direct acoustic path between sources and sensors, and linearity of the transmission channel model is implied with all convolutive mixture models. Finally, some popular algorithms are actually aiming at direction of arrival estimation, which always requires knowledge on the microphone array geometry. In essence, blind algorithms are wellinformed in many respects and are blind only with regard to a few, although decisive, properties.© ASA


FAU Authors / FAU Editors

Kellermann, Walter Prof. Dr.-Ing.
Professur für Nachrichtentechnik


How to cite

APA:
Kellermann, W. (2006). How blind are we in Blind Signal Processing? In Proceedings of the 4th Joint Meeting of Acoustical Society of America and Acoustical Society of Japan (pp. 3046).

MLA:
Kellermann, Walter. "How blind are we in Blind Signal Processing?" Proceedings of the 4th Joint Meeting of Acoustical Society of America and Acoustical Society of Japan 2006. 3046.

BibTeX: 

Last updated on 2019-29-05 at 22:10