Prediction of Channel Excess Attenuation for Satellite Communication Systems at Q-Band Using Artificial Neural Network

This letter proposes the use of an artificial neural network (ANN) for estimating the fading of a Q-band (39.402 GHz) satellite channel exploiting the knowledge of its previous state, as well as the present weather conditions. The ANN is trained using weather data and propagation measurements at the Q-band obtained during a period of nine months by the Aldo Paraboni receivers of RAL Space, Chilbolton, Hampshire, U.K. Subsequently, the estimation obtained by the ANN is compared with actual propagation measurements on data obtained over a period of three months. Statistical analysis demonstrates an agreement between the ANN estimation and the measurement within a 1 dB range with a probability exceeding 98.8%. The significance of this letter lies with the opportunities it raises to deliver real-time fading estimations using low-cost weather sensors combined with feedback on the channel state from the return link, which can be used in the deployment of propagation impairment mitigation techniques.