Coverage and Rate Analysis of Mega-Constellations Under Generalized Serving Satellite Selection

The dream of having ubiquitous and high-capacity connectivity is coming true by emerging low Earth orbit (LEO) Internet constellations through several commercial plans, e.g., Starlink, Telesat, and Oneweb. The analytical understanding of these networks is crucial for accurate network assessment and, consequently, acceleration in their design and development. In this paper, we derive the coverage probability and the data rate of a massive LEO network under arbitrarily distributed fading and shadowing. The conventional user association techniques, based on the shortest distance between the ground terminal and the satellite, result in a suboptimal performance of the network since the signal from the nearest server may be subject to severe shadowing due the blockage by nearby obstacles surrounding the ground terminal. Thus, we take into account the effect of shadowing on the serving satellite selection by assigning the ground terminal to the satellite which provides the highest signal-to-noise ratio at the terminal's place, resulting in a more generalized association technique, namely the best server policy (BSP). To maintain tractability of our derivations and consider the latitude-dependent distribution of satellites, we model the satellites as a nonhomogeneous Poisson point process. The numerical results reveal that implementing the BSP for serving satellite selection leads to significantly better performance compared to the conventional nearest server policy (NSP).