A family of gradient-based independent component analysis techniques for dynamic channel environment: an overview
Thomas Yang · 2009
This paper provides an overview of a family of gradient-based Independent Component Analysis (ICA) techniques. The advantage of these algorithms is the capability to perform blind source separation in dynamic channel environment, a scenario often encountered in many practical applications. A unified presentation is attempted to coherently introduce both block and sequential algorithms. The inherent connections among these various methods are illustrated, which provides a guideline for the selection of a particular method in practice. Key-Words: independent component analysis, gradient method, blind source separation, dynamic environment