Individual Communication Transmitter Identification Using Support Vector Machines with Kernels for Polyspectrum

Polyspectral feature extraction is considered to be a potential method for individual communication transmitter identification. However, the curse of dimensionality caused by higher orders of the features restrains the efficiency of classification. A new method using support vector machine with kernels of polyspectra is present for classification of individual transmitters. The result of experiments on FM and AM individual transmitters shows that the number of support vectors is lower than which using conventional kernel functions, and it can achieve better classification rate.