A short-term rain-induced attenuation model for satellite link quality prediction

This paper proposes a novel but simple satellite link quality prediction based on the relationship between rainfall rate and satellite signal attenuation. After quantifying their relationship statistically, we describe their contemporaneous variations using a linear dynamical system response model. The expectation-maximization (EM) algorithm is used to obtain the maximum likelihood estimates of the model parameters. Subsequently, the communication link quality is predicted by thresholding the estimated short-term signal attenuation level obtained from the Expectation step of the EM algorithm. The prediction performance of the proposed method is compared against the measured beacon signals from Thaicom-4 (IPSTAR) satellite during rainfalls. Using solely the observed rainfall rates from a tipping bucket rain gauge, the algorithm is able to decently predict the satellite link quality. We believe the proposed model can be modified as a predictive tool for real-time link unavailability monitoring of satellite communication system.