Methods and algorithms for decomposition of signal mixture I. Application of decorrelation and second order statistics

The problem of signal mixture decomposition (with reconstruction of the form of its components) is analyzed with no information concerning the proportions and type of mixing. The available measurement information enters the IMS from its sensors. Methods and algorithm structures (necessary for mixture decomposition) are studied. The algorithms are based on decorrelation of multi-dimension signal, covariation matrixes decomposition, and on using the neural network.