B0590
Title: On complex valued time series ICA
Authors: Pauliina Ilmonen - Aalto University School of Science (Finland) [presenting]
Niko Lietzen - Aalto University School of Science (Finland)
Abstract: In the independent component (IC) model, the elements of a $p-$variate random vector are assumed to be linear combinations of the elements of an unobservable $p-$variate vector with mutually independent components. In independent component analysis (ICA) the aim is to recover the independent components by estimating an unmixing matrix that transforms the observed $p-$variate vector to the independent components. Complex random signals play an increasingly important role in the field of ICA. The complex IC model is used for example in magnetic resonance imaging or antenna array signal processing for wireless communications and radar applications. We consider complex valued time series ICA, in particular, we examine the unmixing matrix estimates that are based on simultaneous use of two complex values autocovariance functionals with different lags. We thus extend the well-known AMUSE algorithm for complex valued variables and we also examine the asymptotic behavior of the obtained complex valued unmixing matrix estimates.
