Signals with underlying Markovian properties and their use in modeling attenuation in Ku- and Ka-band mobile satellite transmissions.

Summary The growing demand for broadband satellite services and the congestion of systems operating at traditional frequencies are driving the development of new systems at transmission bands above 10 GHz. However, the behavior of the LMSC (Land Mobile Satellite Channel) is not well known and needs to take into account atmospheric disturbances such as rain. The mobility of antennas for transmission and reception is also part of the problems of satellite services. To design new systems and methods of compensating for losses by dynamic adaptation, models of the channel and rainfall are then necessary. Our work brings first an analysis of the satellite propagation channel at Ku and Ka frequencies by studying on the one hand the normalized received power and on the other hand the precipitation rate time series. We then propose models as well as methods to extract their parameters for these two types of signals. Two approaches, based on MCMC (Monte Carlo Markov Chain) tools, allow a segmentation of the normalized power of the channel as well as the extraction of the parameters of the underlying hidden Markov model. A procedure for evaluating a two-level Markov chain to model the precipitation rate signal is also described. Finally the developed methods are applied to experimental data and provide Markov models of the satellite channel power signal and precipitation rates. The comparison of first and second order statistics between the models and the measurements attests to their quality.