A Fast Asymptotically Efficient Algorithm for Blind Separation of a Linear Mixture of Block-Wise Stationary Autoregressive Processes

0324169 - ÚTIA 2009 RIV TW eng C - Conference Paper (international conference)
Tichavský, Petr - Yeredor, A. - Koldovský, Zbyněk
A Fast Asymptotically Efficient Algorithm for Blind Separation of a Linear Mixture of Block-Wise Stationary Autoregressive Processes.
[Rychlý asymptoticky eficientní algoritmus pro slepou separaci lineární směsi po blocích stacionárních autoregresních procesů.]
Proceedings of 34th IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Taipei: IEEE, 2009, s. 3133-3136. ISBN 978-1-4244-2354-5.
[IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2009 /34./. Taipei (TW), 19.04.2009-24.04.2009]
R&D Projects: GA MŠMT 1M0572
Grant - others:GA ČR(CZ) GP102/07/P384
Program: GP
Institutional research plan: CEZ:AV0Z10750506
Keywords : Approximate joint diagonalization * blind source separation, * autoregressive processes * second-order statistics
Subject RIV: BB - Applied Statistics, Operational Research
http://library.utia.cas.cz/separaty/2009/SI/tichavsky-a fast asymptotically efficient algorithm for blind separation of a linear mixture of block-wise stationary autoregressive processes.pdf

We propose a novel blind source separation algorithm called Block AutoRegressive Blind Identification (BARBI). The algorithm is asymptotically efficient in separation of instantaneous linear mixtures of blockwise stationary Gaussian autoregressive processes. A novel closed-form formula is derived for a Cramer Rao lower bound on elements of the corresponding Interference-to-Signal Ratio (ISR) matrix. This theoretical ISR matrix can serve as an estimate of the separation performance on the particular data. In simulations, the algorithm is shown to be applicable in blind separation of a linear mixture of speech signals.

V práci je navržen nový algoritmus pro slepou separaci signálů. Algoritmus je optimalizován pro separaci po blocích stacionárních procesů, a pokud separované procesz mají Gaussovské rozložení, je též asymptoticky eficientní. Algoritmus je testován na separaci lineární směsi řečových signálů.
Permanent Link: http://hdl.handle.net/11104/0171930