Blind Source Separation Based on ICA

Analysis (ICA) is a new technique to statistically extract independent components from the observed multidimensional mixture of data. Many successful examples of ICA application in the filed of biomedical signal processing are reported recently. In this paper, a steepest descent algorithm of ICA is proposed. A pre-whiten procedure is performed to de-correlating the sensor (mixed) signals before extracting vector. The proposed method is tested with simulated signals and clinical EEG