Research on the order and specific component of the output signals of independent component analysis
Zhenbing Zhao, Qiang Gao, Liyan Huang · 2008
Independent Component Analysis (ICA) is a new signal processing technique that can recover source signals from the observed mixtures. In some applications of ICA, it is desired that the order of the recovered source signals can be predicted and the component which we are interested in can be extracted separately. But the order of recovered source signals is unpredictable. Facing this problem, a new algorithm based on ICA is presented for determining the order of recovered source signals in this paper. First it gives a brief introduction about the problem of ICA, and then by using the observed vectors and a priori knowledge of components from the first run, it also constructs the linear system of equations statistically and achieves the separating matrix. The results of simulation show that the proposed method can separate the source signals in specified order effectively.