Connectivity of LEO Satellite Mega Constellations: An Application of Percolation Theory on a Sphere

Hao Lin, Mustafa A. Kishk, Mohamed-Slim Alouini

Introduction

The transition toward the 6G telecommunications era is characterized by the deployment of massive Low Earth Orbit (LEO) satellite constellations. Unlike traditional geostationary satellites, these mega-constellations—such as SpaceX’s Starlink, Amazon’s Kuiper, and OneWeb—rely on thousands of small satellites to provide global internet coverage. While existing research often utilizes stochastic geometry to evaluate average coverage at specific geographic points, it frequently overlooks the systemic property of "global connectivity." This paper addresses this gap by providing the first formal application of percolation theory to a spherical surface to identify the critical thresholds required for a constellation to form a continuous, giant service area.

The Problem of Global Connectivity

In network science, percolation describes the phase transition where isolated clusters within a system merge to form a "giant component" that spans the entire network. On a flat 2D plane, these thresholds are well-documented. However, the Earth’s surface is a closed sphere, introducing geometric complexities that traditional models fail to capture. The central question is: at what point does a collection of individual satellite coverage footprints transition from a series of "patchy" islands into a unified, globally connected network?

Methodology: Bridging 3D Geometry and 2D Percolation

To analyze percolation on a sphere, the authors developed a mathematical framework that maps 3D spherical geometry onto a 2D plane. The methodology involves several key steps:

  • Defining Spherical Percolation: The authors define percolation as the probability of establishing a continuous coverage area that connects two antipodal points (e.g., the North and South Poles).
  • Stereographic Projection: To leverage established 2D percolation proofs, the Earth’s surface is projected onto a 2D plane tangent to the South Pole. The authors rigorously prove that if coverage areas are connected on the sphere, their corresponding projections remain connected on the plane.
  • Hexagonal Lattice Mapping: By utilizing a hexagonal lattice on the projected plane, the researchers established upper and lower bounds for the "critical number" of satellites.
  • Phase Transition Analysis: The study demonstrates the existence of a distinct phase transition—a threshold where the probability of global connectivity shifts abruptly from zero to near-certainty.

Key Insights: The 50% Rule and the Critical Number

A significant finding of this research is the "50% Rule." The authors demonstrate that continuous percolation on a sphere occurs when the probability of any single point being covered ($p_{cov}$) reaches exactly 0.5. Essentially, if more than half of the Earth’s surface is covered by satellite footprints, a giant component of connectivity is almost guaranteed to exist.

Furthermore, the authors derived an elegant closed-form expression for the critical number of satellites ($N_c$) required to reach this phase transition:

$$N_c = \frac{\ln 2}{\ln 2 - \ln(1 + \cos \gamma)}$$

In this formula, $\gamma$ represents the coverage angle, which is determined by the satellite's altitude ($h$) and the maximum slant range ($d_m$). This suggests that connectivity is not merely a function of the number of satellites, but a delicate balance of orbital geometry.

Real-World Application and Results

The theoretical model was validated using the parameters of three major LEO systems. The results provide concrete thresholds for achieving a "giant" continuous service area:

  • Starlink: Operating at an altitude of 550 km with a 5.2° coverage angle, it requires at least 340 satellites to reach the percolation threshold.
  • OneWeb: At 1200 km altitude and a 6.14° coverage angle, the threshold is approximately 240 satellites.
  • Project Kuiper: At 610 km altitude and a 6.58° coverage angle, the threshold is approximately 200 satellites.

Numerical simulations closely matched these theoretical critical values, confirming that the 2D projection method is a robust tool for analyzing 3D spherical networks. The data also highlights the impact of altitude: while higher altitudes significantly reduce the number of satellites needed for global connectivity by expanding individual footprints, they also introduce higher latency.

Limitations and Future Research

While this paper provides a foundational framework, it acknowledges several limitations:

  • Uniform Distribution: The model assumes a Binomial Point Process (BPP) where satellites are uniformly distributed. Real-world constellations follow specific orbital shells, leading to higher densities near the poles.
  • Temporal Dynamics: The analysis provides a static snapshot. Future work must account for the high-speed orbital propagation of satellites, which causes the giant component to shift dynamically over time.
  • Capacity and Routing: Geometric connectivity does not equate to network capacity. Future studies should integrate bandwidth constraints and routing algorithms to determine if a connected component can actually support user traffic.

By establishing the first formal application of percolation theory to a spherical surface, this research provides a critical benchmark for the design and deployment of future 6G satellite infrastructures.

Summary updated

BibTeX
@article{lin2025connectivity,
  author = {Hao Lin and Mustafa A. Kishk and Mohamed-Slim Alouini},
  title = {{Connectivity of LEO Satellite Mega Constellations: An Application of Percolation Theory on a Sphere}},
  journal = {IEEE Trans. Aerosp. Electron. Syst.},
  year = {2025},
  doi = {10.1109/taes.2025.3553461},
}
Figure