In the recent years, Blind source separation by independent component analysis has gained popularity in the recent years because of its applications in signal processing such as in speech recognition systems, telecommunications and medical signal processing such as electroencephalogram (EEG), electrocardiogram (ECG). Many methods have been developed to implement the blind source separation of which independent component analysis (ica) has emerged as a winner. Ica estimates the independent components from the mixed signal can be broadly defined in two ways i.e. minimizing the mutual information or maximizing the non-gaussianity.;Representation of mixed signal as mixture of independent signals is the instantaneous mixing model, practically the signal undergoes multipath distortion throughout the channel length, and so convoluted mixing model is used to obtain the mixed signal. Separation of the mixed signal is achieved by an algorithm, RADICAL (Robust, Accurate, and Direct Independent Component Analysis).;The objective of the thesis is to implement a new algorithm that has been is developed recently and being called as RADICAL ica algorithm along with the entropy estimation concepts from the statistics literature. Objective function is developed by considering the Kullback-Leibler divergence between the joint distribution and product of marginal distributions, RADICAL algorithm is implemented along with the modified version of the entropy estimator developed by vasicek to optimize the objective function and develop an un-mixing matrix to get back the estimated independent components.
展开▼
