Astrolabe: Modeling RTT Variability in LEO Networks.

Networking practitioners heavily rely on intuitive models of the behavior of networks when designing and analyzing protocols and algorithms. However, there is still a lack of such intuitive models of the behavior of LEO satellite networks, hindering innovation. In this paper, we provide a first step towards improving the intuitive understanding of the behavior of LEO satellite networks. In particular, we focus on developing a model that captures the RTT variability exhibited by such networks. We rely on simple and intuitive calculations instead of expensive simulations. To capture the high RTT variability exhibited by satellite networks, we estimate lower and upper bounds for the RTT between a pair of ground stations. We introduce Astrolabe, a novel approach that achieves accurate bounds, with a median lower bound within 1.15X the actual lowest RTT and a median upper bound within 2X the highest RTT in a few seconds instead of hours required by simulations or measurements.