Demodulation of Low SNR QPSK Signal Based on CNN

Abstract

In order to overcome the problems of high bit error rate of QPSK demodulation under low signal-to-noise ratio in traditional modulation pattern recognition methods, a new QPSK demodulation method under the condition of low signal-to-noise ratio based on convolutional neural network is proposed. This method inputs the IQ signal into the denoise network, and then the convolution neural network is used for demodulation. The loss function is the cross entropy of the demodulated symbol sequence and the original symbol sequence multiplied by the transmitter RF signal and the denoise of the receiver. On the basis of the above, through this method, the bit error rate of QPSK demodulation when the signal-to-noise ratio is in a relatively low state can be effectively reduced. Simulation results show that this method can effectively reduce the bit error rate under the condition of low signal-to-noise ratio. When the SNR is -20dB, the average bit error rate is 6.98%, which is 38.85% lower than traditional coherent demodulation.