Time Domain Optimization Techniques for Blind Separation of Non-stationary Convolutive Mixed Signals

I. Russell, A. Mertins, and J. Xi (Australia)


Blind source separation, Global optimization, Joint diagonalization, multivariate optimization, Newton method,Steepest gradient descent


This paper aims to solve the problem of Blind Signal Sep aration (BSS) in a convolutive environment based on out put correlation matrix diagonalization. Firstly an extension of the closed form gradient and Newton methods used by Joho and Rahbar [1] is developed which encapsulates the more difficult convolutive mixing case. This extension is completely in the time domain and thus avoids the inherent permutation problem associated with frequency domain ap proaches. We also compare the performance of three com monly used algorithms including Gradient, Newton and global optimization algorithms in terms of their conver gence behavior and separation performance in the instan taneous case and then the convolutive case.

Important Links:

Go Back