Neural Network Demodulator for Frequency Shift Keying

In this paper, we propose a novel artificial neural network (ANN) demodulator to demodulate FSK signal. It has some important features compared with conventional method. Firstly, the anti-interference ability of ANN demodulator is better than that in traditional way. In traditional receiver, there must be a band-pass filter (BPF) to filter out-of-band noise; however, in ANN demodulator, the signal is never processed by any filter beforehand. Secondly, the ANN demodulator has necessary function modules for FSK demodulation, including BPF, pulse former, and decoder; however these modules neednpsilat be designed separately but are acquired by ANN demodulatorpsilas self-learning. Thirdly, its process of demodulating signal is concurrent, so the operating speed is rapider than that in conventional way. Finally, ANN demodulator is all-purpose system, that means the same system can demodulate different signals including ASK signal, FSK signal, etc. The effectiveness is proved by the simulation result of MATLAB.