Recognition of digital modulation signals based on high order cumulants and support vector machines

The paper presents a new algorithm of digital modulation signals based on high order cumulants (HOC) and support vector machines (SVM). The parameters which are picked up from the signals' fourth order and sixth order cumulants are used as the classification feature vectors. Using SVM based on binary tree as classifiers, recognition of the 2ASK, 4ASK, QPSK, 2FSK and 4FSK signals is efficient. The computer simulation results justify that the success rate is over 97.5% at SNR = 10 dB.