Modeling communications in low-earth-orbit satellite networks

This thesis studies communications in low-earth-orbit satellite networks. It develops simple analytical models of networks formed by interconnecting the satellites. Then it uses the models to study efficient communication schemes in these networks. The use of wireless communications has increased rapidly over the past years. Since the cellular systems are often cost prohibitive in sparsely populated areas, several consortia are involved in building alternative systems that use low or medium orbit satellites. Compared to geostationary orbits, the low orbits achieve smaller delay but result in smaller coverage areas requiring a large number of satellites. The satellites are in constant motion, which results in high variability in the networks. This thesis studies the communication networks that result from interconnecting low-earth-orbit satellites in inclined orbits. The first part of the thesis focuses on modeling two aspects of the networks—the network topology and the transmission delay. It provides an extensive numerical study of the impact of various parameter on the delay between both directly and indirectly connected satellites. It shows that, within a certain class of topologies, a new network topology called a skewed torus minimizes the delay within the network. Due to the complex formula for the exact value of intersatellite delay, it also introduces two approximations called the linear approximation and the constant approximation. It discusses the distinction between the links connecting satellites on the same orbit and on different orbits, and proposes the use of the constant approximation for interorbital links resulting in a simple two-uniform model. The second part of the thesis develops efficient communication algorithms for toroidal networks under the two-uniform model. The focus is on all-to-all exchange (gossiping) algorithms. The thesis develops algorithms under two transmission-cost models: the constant-cost model that considers propagation delay only and the linear-cost model that considers both propagation delay and data rate. An algorithm that utilizes the overlap between propagation delay and transmission time on different links is developed for the linear-cost model. The algorithm improves the time of the best-known algorithm in the special case of a one-uniform regular torus (all links have the same parameters).