Automatic modulation classification by support vector machines

Automatic classification of analog and digital modulation signals plays an important role in communication applications such as an intelligent demodulator, interference identification and monitoring, so many investigations have been carried out in the past. Support Vector Machines (SVMs) maps inputs vectors nonlinearly into a high dimensional feature space and constructs the optimum separating hyperplane in space to realize signal classification. In this paper, a new method based on SVM for classifying AM, FM, BFSK, BPSK, USB and LSB is proposed. The classification results for real communication signals using SVMs are given. Compared with radial basis function neural network (RBFNN) method, the method can classify these signals well, and the correct classification rates are above 82%.