A Novel Wireless Channel Model with Multiply Feed-forward Neural Network

According to the mechanism of feed-forward neural networks (FNN), this paper describes a wireless communication channels model based on multilayer FNN. Utilizing the excellent learning characteristics of neural networks (NN), give a modeling method with improved back propagation (BP) algorithm and the simulation results, analyze the existed problems and the settling means as well. Simultaneously, simulation results show that the FNN model can track the time-varying characteristic of nonstationary wireless channels very well. When applied to practical wireless communication channels, several unique merits of this method are also demonstrated.