Neural Network for Demodulating the Output Signals of Nonlinear Systems with Memory

This paper proposes a neural network (NN) approach for demodulating output signals of a nonlinear channel with memory. The feed-forward neural network is trained to learn the appropriate mapping between nonlinear input patterns and source bits. The simulation results provide some evidence that neural networks can learn the effect of nonlinear channels with memory and demodulate the output signal of such a channel effectively. Further more, we observe that the neural network trained on the partial nonlinearity of the channel can demodulate the output signals of the channel with complete nonlinearity, which can be used to reduce the training complexity.