Penemuan Sinyal Asli dengan Metode Independent Component Analysis pada Sinyal Tercampur Tunggal

Abstract

Independent Component Analysis (ICA) is a popular technique to find the components forming mixed signals on the assumption that at least one of the constituent components is non-gaussian. ICA technique generally requires that at least the same number of mixed signals with its constituent components. So in case one mixed signal is lost or damaged, the original signal can not be found. The authors propose a new method using ICA for a single mixed signal. With these new methods, techniques ICA is able to find the original signal of a mixed signal. This method of recording one of the mixed signals, which are not damaged, with another signal forming mixed signals. This method requires two things. First, Another original signal is known. Second, the signal recording conditions are always the same in any condition. In computing, signal recording conditions represented as a matrix mixer. Of the mixed signal that this is the second time the ICA technique is used to find the original signal. Based on the experiments, the new method managed to find the original signal forming mixed signals. These results open the possibility of solving opportunities separation of data in a single mixed the data for cases that are more specific example, the disposal of a single signal noise, the detection of false images and so on.