Recognition of digital modulation signals based on hybrid three-order restricted Boltzmann machine

A restricted Boltzmann machine (RBM) is a type of neural network model that has a two-layer architecture. Any two units in the different layers are connected and in the same layer are not connected. Hybrid 3-order RBM (H3RBM) is used as a classifier without setting a threshold and its training time is very short. One feature parameter extracted from four-order and eight-order cumulants is used to recognize 2ASK, 4ASK, QPSK, 8PSK, MFSK, 16QAM and it can suppress white Gaussian noise. Simulation results show that the feature parameter is unaffected by white Gaussian noise. The classification method requires only one parameter, and has high recognition rates and low complexity.