Automatic digital modulation identification basing on decision method and cumulants

Automatic modulation identification through digital signal processing has many applications in wireless communication systems. But most of signal classification methods can only recognize a few kinds of signals and they request SNR should be beyond 10 dB. Based on the method presented by Asoke K. Nandi and E. E. Azzouz, this paper adds some high-order cumulants, and uses the RBF neural network to classify 2ASK, 4ASK, 8ASK, 2PSK, 4PSK, 8PSK, 2FSK, 4FSK, 8FSK, 16QAM, 64QAM, these eleven kinds of digital modulated signals in low SNR to -5 dB. The simulation results indicate that this improved method can get good results.