Blind separation of convolved sources using the independent component analysis and information maximization approach

Independent Component Analysis (ICA) is very closely related to the method called blind source separation (BSS) or blind signal separation. In Independent Component Analysis (ICA) components are assumed statistically independent which we call independent source signal. In our thesis we have considered only noiseless ICA case. In a number of real-world signal processing applications, signals from various independent sources may get distorted by environmental factors that can be represented as convolutive mixtures of original signals received at the sensors. In this thesis, the effects of environmental factors and modeling assumptions on the performance capabilities of independent component analysis-based techniques are investigated. The so-called blind source separation feedback network architecture that is capable of coping with convolutive mixtures of sources is derived using Bell and Sejnowski's information maximization principle