Abstract:In this paper we present a novel approach to generalized dynamic independent component (DICA) analysis, especially multichannel blind deconvolution problems. Assuming that separating (demixing) dynamic system is described by stable linear state-space models, we decompose the blind deconvolution problem into two stages: blind separation/deconvolution and equalization. We first develop a learning algorithm to update the parameters in output equation of the demixing model. In the separation/deconvolution stage we separate mixed signals into mutually independent signals. In the second stage we apply effective single channel equalisation algorithms in order to estimate original source signals. In this paper we also discuss briefly nonlinear state space models for blind and semi-blind separation of linearly/nonlinearly mixed and filtered (convolved) independent sources. For nonlinear mixture Hyper Radial Basis Function (HRBF) neural network proposed by Poggio and Girosi is employed and associated supervising - unsupervising learning rules for its parameters are proposed. Theoretic analysis and computer simulations confirmed validity and high effectiveness of the proposed method.

Blind Identification and Signal Separation of Two-Input Two-Output FIR Systems

Rapid Second-Older Blind Source Separation Algorithm for Vibration Signals

Blind Channel Identification with Colored Sources by Exploiting Properties of Companion Matrices.

Some further results on blind identification of MIMO FIR channels via second-order statistics

Blind MIMO FIR Channel Identification by Exploiting Channel Order Disparity

Bi-iterative least squares algorithms for blind channel identification and equalization with second-order statistics

Blind Separation Using Second Order Statistics for Non-Stationary Signals

Underdetermined Blind Separation of Overlapped Signals of the Same Frequency

An algebraic method for blind FIR channel identification with cumulant matrix decomposition

Two-stage Blind Deconvolution Using State-space Models.

On the use of signal subspace information in blind FIR channel estimation

Blind Identification of Nonlinear Finite Impulse Response Volterra Channels

A Cumulant Subspace Projection Method For Blind Mimo Fir Identification

Blind Identification of SIMO Wiener Systems based on Kernel Canonical Correlation Analysis

A new method of blind inverse identification of system

Blind separation of convolutive mixtures based on second order and third order statistics

Second-order statistics method based on total least-squares for blind separation of stationary sound signals

Blind Equalization of SIMO FIR Channels Driven by Colored Signals with Unknown Statistics.

Adaptive identification of nonminimum phase FIR systems

Blind Two-Dimensional Super Resolution in Multiple Input Single Output Linear Systems

Blind separation and filtering using state space models