Adaptive multichannel blind deconvolution using state-space models

Independent component analysis (ICA) and related problems of blind source separation (BSS) and multichannel blind deconvolution (MBD) problems have recently gained much interest due to many applications in biomedical signal processing, wireless communications and geophysics. In this paper both linear and nonlinear state space models for blind and semi-blind deconvolution are proposed. New unsupervised adaptive learning algorithms performing extended linear multichannel blind deconvolution are developed. For a nonlinear mixture, a hyper radial basis function (HRBF) neural network is employed and associated supervised-unsupervised learning rules for its parameters are developed. Computer simulation experiments confirm the validity and performance of the developed models and associated learning algorithms