Gradient and Fixed-Point Complex ICA Algorithms Based on Kurtosis Maximization

Abstract

We present two algorithms for independent component analysis of complex-valued signals based on the maximization of absolute value of kurtosis and establish their properties. Both the algorithm derivation and the analysis are carried out directly in the complex domain, without the use of complex-to-real mappings as the cost function satisfies Brandwood's analyticity condition. Simulation results are presented that show the advantages of the new algorithms, especially when the number of sources in the mixture increases.

Keywords

This publication has 15 references indexed in Scilit:

ICA by Maximization of Nongaussianity using Complex Functions
Published by Institute of Electrical and Electronics Engineers (IEEE) ,2006
Complex independent component analysis of frequency-domain electroencephalographic data
Neural Networks, 2003
Monotonic convergence of fixed-point algorithms for ICA
IEEE Transactions on Neural Networks, 2003
Approximation by Fully Complex Multilayer Perceptrons
Neural Computation, 2003
Polar coordinate based nonlinear function for frequency-domain blind source separation
Published by Institute of Electrical and Electronics Engineers (IEEE) ,2002
Globally convergent blind source separation based on a multiuser kurtosis maximization criterion
IEEE Transactions on Signal Processing, 2000
A Fast Fixed-Point Algorithm for Independent Component Analysis of Complex Valued Signals
International Journal of Neural Systems, 2000
A Fast Fixed-Point Algorithm for Independent Component Analysis
Neural Computation, 1997
The constrained total least squares technique and its applications to harmonic superresolution
IEEE Transactions on Signal Processing, 1991
New criteria for blind deconvolution of nonminimum phase systems (channels)
IEEE Transactions on Information Theory, 1990

Cited by 15 articles