This lecture is a research result concerning the application of Neural Networks to Communication Problems. Mobile Communication Channels present two difficult problems of System theory: time-varying transfer functions (due to mobility) and Non Linearity (mainly due to satellite segments in the channel). Quite a few techniques are able to cope with this kind of situation, mainly when real time processing constraints are added. This lectures intends to show that Neural Networks are able to provide very flexible and efficient solutions to two most important classes of problems in the field: the mobile channel identification, aiming at providing a NN channel model, and the equalization problem of non linear, non stationary channels. After a brief reminder of Neural Networks main structures and properties, an outline of questions raised by Mobile Communications Channels (time-varying, multipath, Doppler, on board satellite Non Linear Amplifiers) are presented. The papers concludes with examples of Identification and Equalization of mobile non linear channels.