Neural network modeling of land mobile radio channels with nonlinear transmit amplifiers

Mobile radio transmission is characterized by the use of a high-power amplifier due to severe channel fading. In addition, it is a nonstationary transmission because of the time-varying channel, since the transmitter is a mobile user. Neuronal systems present the advantage of being nonlinear and have a good ability for apprenticeship. That justifies the approach of neuronal modeling that we propose in this paper. We propose a neuronal structure for the modeling of nonlinear multipath mobile radio channel. The model weights are updated using the backpropagation algorithm; except the time-varying block modeling the multipath effect, which is adapted by the recursive least square algorithm allowing nonstationarity tracking.