Demonstration of an observation tool to evaluate the performance of ICA technique

The motivation behind the Blind source separation is to separate the mixed sources, in particular for the blind case where both the sources and mixing process are unknown and if desirable to recover all the sources from the mixtures,. The paper reveals the blind source separation techniques, the mixing environment probably occurs with signals. The present work compares the performance of the Principal Component Analysis (PCA) technique and Independent component analysis (ICA). The demixing process used is based on the maximization of Kurtosis. The extent of demixing is assessed from the strength of the scaled version of off diagonal elements in the correlation matrix of demixed output. The Matlab simulation supplemented by plots, scatter, and bar diagrams between signal separated brings out effectively the superiority in the performance of the maximization of Kurtosis for source separation.