Application of Monte Carlo Markov chain to determination of hidden Markov model for mobile satellite channels

The fluctuations of mobile satellite channels are usually modelled by Markov chains. Existing models postulate the number of states, and their associated distributions, based on physical considerations. This produces good models but ones which are not convenient in different contexts. We focus on the methodology of extraction of hidden Markov models (HMM) from experimental data to describe the time fluctuations of received power in a mobile satellite service (MSS) context. It is based on an MCMC (Monte Carlo Markov chain) method associated with a k-means classification. Its complexity is reduced when compared to the traditional MCMC method. Contrary to existing detection methods, the only assumption is the HMM states number and it enables an accurate estimation of the HMM parameters and of the transitions location between states of the model.