Markov approximation for multilayered selection in satellite network

Abstract

A satellite network system comprises three layers of satellites: LEO (Low Earth Orbit), MEO (Middle Earth Orbit) and GEO (Geostationary Orbit). In the system, users can choose a layer according to their demands, including QoS (Quality of Service), congestion, energy cost, etc. The utility that users gain will change when they access satellites in different layers. The mobility of satellites in the LEO and MEO leads to frequent handover among satellites in the same layer. These characteristics of multi-layered satellite networks make it possible for us to exploit the optimal distribution of users, which will maximize the utility of the entire satellite network. While the proposed problem is an NP-hard problem, we analyze the system based on the Markov chain and use the Markov approximation to approach the maximum utility. In addition, we use the count down and select algorithm to implement the process of Markov chain. The simulation results validate the convergence of the Markov approximation. In addition, the gap between the approximate value and optimal values decreases with an increase in ß, which is a positive constant in Markov formulation, according to the simulation results.