Performance Analysis for Data Compression Based Signal Classification Methods
In this paper, the authors consider the blind signal classification problems. In the considered scenarios, a sequence of random signal samples X1, X2,. .., XN is observed, where each signal sample is a real number or a vector in a finite dimensional space. It is assumed that the N signal samples are generated by J information sources with different statistical properties. However, it is unknown from which information source each signal sample Xn is emitted. The blind signal classification problems denote the problems of estimating the membership of each signal sample to the information sources.