Discrete wavelet neural network group system for digital modulation recognition

In this paper, neural network group is proposed to improve the performance of discrete wavelet neural network group (DWNNG) for automatic digital modulation recognition (ADMR). The neural network group can reduce the complexity of only one neural network. The neural network can be easier to be trained with low errors by using neural network group. The problem of local optimum can be improved. And the system can recognize more modulations. The experiment indicates that it can recognize 2ASK, 4ASK, 8ASK, 2FSK, 4FSK, 8FSK, 2PSK, 4PSK, 8PSK, 4QAM and 16QAM. The mean correct recognition rates were obtained 95.7% by using neural network group substitute for single one.