Abstract
This research introduces a rigorous mathematical framework for analyzing the "localizability" of next-generation hybrid satellite navigation systems. As global navigation satellite systems (GNSS) evolve toward "mega-constellations" integrating Low Earth Orbit (LEO) and Medium Earth Orbit (MEO) satellites, traditional simulation-based performance evaluations become computationally prohibitive. This paper presents the first stochastic geometry-based framework to analyze these hybrid networks, introducing a novel Doubly Stochastic Binomial Point Process (DSBPP) to accurately model structured MEO orbits. The derived analytical formulas predict the probability of a ground user connecting to sufficient satellites for precise positioning, accounting for interference, bidirectional beam alignment, and complex orbital geometries.
Introduction
Global Navigation Satellite Systems (GNSS), such as GPS, have traditionally relied on Medium Earth Orbit (MEO) satellites. While reliable, these systems face limitations in high-precision, low-latency applications. Integrating Low Earth Orbit (LEO) satellites—which are more numerous and closer to Earth—offers a promising solution for enhanced performance. However, evaluating these hybrid networks is challenging; traditional "brute force" Monte Carlo simulations do not scale well with thousands of LEO satellites. Furthermore, standard random models like the Poisson Point Process (PPP) fail to capture the highly structured, deterministic nature of MEO orbits. This research addresses this gap by providing a low-complexity analytical model that captures the unique geometry of satellite constellations.
Methodology
The researchers utilized Stochastic Geometry (SG) to model the spatial distribution of satellites. Key methodological innovations include:
- The DSBPP Model: To accurately represent MEO satellites, the authors proposed the Doubly Stochastic Binomial Point Process. This model treats orbits as one random process and the satellites within those orbits as another, effectively capturing the "lines in the sky" characteristic of structured constellations.
- K-Localizability Probability: A new metric was defined to calculate the probability that at least $K$ satellites are both visible (availability) and provide a sufficient Signal-to-Interference-plus-Noise Ratio (SINR) for positioning.
- Bidirectional Beam Alignment: The model accounts for directional antennas on both the satellites and the ground target, which are essential for mitigating interference in dense networks.
- Realistic Channel Modeling: The framework incorporates Shadowed Rician Fading to account for atmospheric and environmental signal degradation.
Key Insights and Results
The analytical formulas derived in this study were validated against Monte Carlo simulations, showing near-perfect alignment and proving the SG approach as a reliable, low-complexity alternative to heavy simulations. Several critical insights emerged:
- The 150-to-1 Rule: The study identified a complementary relationship where removing a single MEO satellite requires the addition of approximately 150 LEO satellites to maintain the same level of availability.
- The Altitude "Sweet Spot": For LEO satellites, increasing altitude (e.g., from 1,000 km to 2,000 km) improves visibility but also increases interference and path loss. The research identifies a peak performance point (approximately 1,500 km) where these factors are optimally balanced.
- Interference Bottlenecks: While adding LEO satellites initially improves availability, dense networks eventually face a "localizability" decrease as cumulative interference begins to drown out positioning signals.
- Performance Gains: Adding 2,000 LEO satellites to a sparse 12-satellite MEO network can increase the probability of having six usable signals by up to 60%.
Limitations and Future Work
While providing a robust framework, the study notes certain simplifications. The current model primarily considers the impact of the closest interfering satellite to maintain mathematical tractability. Future work will aim to incorporate more exhaustive interference models and complex, adaptive phased-array antenna patterns. Additionally, the researchers plan to extend this "localizability" framework (answering if enough satellites are visible) to predict actual Positioning Error in meters using metrics like the Cramér-Rao Lower Bound (CRLB).
Summary updated