Abstract of Thesis presented at COPPE/UFRJ as a partial fulfillment of the requirements for the degree of Doctor of Science (D.Sc.)

Rain Attenuation Prediction in Earth-Space Paths Using Artificial Neural Networks

Gilson Alves de Alencar

March/2003

Advisor:

Luiz Pereira Calôba

Department:

Eletrical Engineering

      With the expansion of satellite communications in the last years, it has being necessary to allocate higher frequecies bands to support the new services. Nevertheless, signals at super high frequencies suffer hard degradation due to rain attenuation. Several phenomenological models have been developed to estimate this attenuation and a great number of studies have been carried out to make the accuracy of these models better. This work proposes two new methods to evaluate the rain attenuation in earth-space paths. First a specific neural network was carefully designed and developed to perform an adequate rain attenuation prediction. The proposed neural model has shown to be able fo evaluate the rain attenuation with good accuracy. At the second part of this work we developed a technique to identify fails in the ITU-R earth-space propagation model and a phenomenological-neural hybrid model was created to adjust it. The hybrid solution has shown to have an excellent accuracy and it seems to be possible to still increase their performance.