The Projection Approximation Subspace Tracking Algorithm Applied to Whitening and Independent Component Analysis in Wireless Communications
Blekinge Institute of Technology, School of Engineering, Department of Signal Processing2005 (English)Report (Other academic)
In Blind Source Separation (BSS) the objective is to extract source signals from their linear mixtures. Algorithms developed for Independent Component Analysis (ICA) have proven useful in the field of BSS. The Projection Approximation Subspace Tracking with Deflation (PASTD) algorithm, originally developed for subspace tracking, has been extended by using a nonlinear cost function so that it may be used for ICA/BSS. Such algorithms most often require the input signals to be white. In this report we extend the PASTD algorithm so that it can be used to whiten signals as a pre-processing step before ICA. The performance of the ICA-algorithm is then evaluated for different choices of whitening algorithms. The algorithms are also evaluated for Binary Phase Shift Keying (BPSK) modulated data over Rayleigh fading channels usually encountered in wireless communications.
Place, publisher, year, edition, pages
Blekinge Tekniska Högskola Forskningsrapport, ISSN 1103-1581 ; 1
Blind signal processing, blind source separation, independent component analysis, nonlinear principal component analysis, projection approximation, subspace tracking, recursive least squares, wireless communications, Rayleigh fading channels, whitening, decorrelation.
Telecommunications Signal Processing
IdentifiersURN: urn:nbn:se:bth-00301Local ID: oai:bth.se:forskinfoE4682615FC263EE2C1256F7E003BFBC2OAI: oai:DiVA.org:bth-00301DiVA: diva2:837276