The impact of LEO satellite hardware delays on LEO-augmented precise point positioning

Zihao Ye, Kan Wang, Chunbo Wei, Beixi Chen, Jinqian Wang, Meifang WuGPS 2025 · DOI

Abstract

Low Earth Orbit (LEO) satellite augmentation is considered a promising solution to enhance Precise Point Positioning (PPP) performance, offering stronger signals and faster geometry changes. Notably, in the absence of a global ground station network tracking LEO navigation signals, LEO-based Positioning, Navigation, and Timing (PNT) applications must depend on orbital and clock products derived from onboard GNSS observations. This operational reliance inherently introduces not only LEO satellite orbital and clock errors, but also hardware biases that distinguish between those of the onboard GNSS receiver contained in the clock estimates and those of the downlinking antenna that are needed by the users. This study investigates the impacts of these potential LEO-induced errors on LEO-augmented multi-GNSS PPP. Code biases are simulated with different offsets and periodic behaviors for a CentiSpace-like LEO satellite constellation of 150 satellites, and real orbital and clock errors from Sentinel-3B are used to derive the real-time LEO satellite orbital and clock errors used for simulations. Results demonstrate that convergence time and positioning accuracy are strongly degraded, sometimes even performing worse than GNSS-only solutions when code biases reach tens of nanoseconds.

Introduction: The "Patience Problem" in PPP

Precise Point Positioning (PPP) is the gold standard for high-accuracy navigation in applications ranging from autonomous agriculture to seismic monitoring. However, traditional GNSS satellites (GPS, Galileo) orbit at high altitudes, leading to slow changes in geometry relative to receivers. This often requires 20–30 minutes for the solution to "converge" to centimeter-level accuracy.

LEO satellites, orbiting much closer to Earth, move rapidly across the sky, theoretically enabling much faster convergence. However, LEO navigation lacks a global ground network for signal calibration. This creates a gap in modeling "code biases"—internal hardware delays between the onboard GNSS receiver and the downlinking antenna. If these delays are not accurately modeled, the rapid movement of LEO satellites may accelerate error injection rather than improving precision.

Methodology and Simulation Design

To evaluate these impacts, researchers utilized a "hardware-in-the-loop" style simulation. The study modeled a 150-satellite CentiSpace-like constellation distributed across two orbital layers to ensure global coverage.

Global ground tracks of the simulated 150-satellite LEO constellation.

Error and Bias Modeling

Rather than relying on idealized mathematical models, the study incorporated real-world orbital and clock error data from the Sentinel-3B satellite to represent realistic LEO behavior. This data provided a baseline for the magnitude and stability of errors expected in operational LEO missions.

Orbital error characteristics (Radial, Along-track, Cross-track) and their stability (Modified Allan Deviation) derived from Sentinel-3B.

The researchers simulated code hardware delays using three scenarios: constant offsets, periodic variations (simulating orbital temperature swings), and a combination of both. Processing was performed using an Extended Kalman Filter (EKF) to integrate simulated LEO signals with real-world GNSS data from 40 Multi-GNSS Experiment (MGEX) stations collected on September 1, 2024.

Key Insights: The Geometry Projection Trap

The study identifies a critical paradox: the fast movement of LEO satellites, which is their primary advantage, also makes them highly sensitive to hardware delays. Because the satellite moves so quickly, a static hardware bias is projected into the position solution at a rapidly changing angle. This prevents the positioning filter from "averaging out" the error, effectively confusing the filter.

Surprisingly, the constant offset in hardware delay proved to be the primary performance killer. While temperature-induced periodic variations exist, their impact on convergence is minor compared to the raw magnitude of uncalibrated biases. There is a threshold where LEO augmentation stops being an "upgrade"; if code biases reach tens of nanoseconds, the LEO-augmented solution can become less accurate than a traditional GNSS-only solution.

Results and Performance Degradation

The simulation revealed a clear "tipping point" where LEO augmentation ceases to be beneficial:

  • Convergence Speedup: In an ideal scenario (no errors), LEO reduces convergence time by 50% to 90%.
  • The Cost of Realism: Introducing realistic orbital and clock errors from Sentinel-3B increased static PPP convergence time from 3.2 minutes to 11.0 minutes.
  • The Impact of Bias: Under high-bias scenarios (40 ns), kinematic positioning convergence time ballooned to 25.8 minutes—nearly negating the benefits of adding LEO satellites.
  • Accuracy Degradation: At the BRUX station, a 40 ns bias caused the vertical (Up) RMS error to increase by 42%, dropping precision significantly compared to GNSS-only solutions.

Conclusion and Future Outlook

While LEO satellites offer a path to near-instantaneous high-precision positioning, uncalibrated hardware delays present a significant hurdle. Future research must focus on developing in-orbit calibration techniques to estimate these biases dynamically without a dense ground station network. Additionally, the study focused primarily on code biases; the impact of uncalibrated phase biases, critical for integer ambiguity resolution, remains a key area for further investigation.

Summary updated

Abstract

Abstract Low Earth Orbit (LEO) satellite augmentation is considered a promising solution to enhance Precise Point Positioning (PPP) performance, offering stronger signals and faster geometry changes. Notably, in the absence of a global ground station network tracking LEO navigation signals, LEO-based Positioning, Navigation, and Timing (PNT) applications must depend on orbital and clock products derived from onboard GNSS observations. This operational reliance inherently introduces not only LEO satellite orbital and clock errors, but also hardware biases that distinguish between those of the onboard GNSS receiver contained in the clock estimates and those of the downlinking antenna that are needed by the users. The on-ground calibration of these biases might not cover their in-orbit variations. This study investigates the impacts of these potential LEO-induced errors on LEO-augmented multi-GNSS PPP. Code biases are simulated with different offsets and periodic behaviors for a CentiSpace-like LEO satellite constellation of 150 satellites, and real orbital and clock errors from Sentinel-3B are used to derive the real-time LEO satellite orbital and clock errors used for simulations of the constellation. Using real GNSS observation data from 40 Multi-GNSS Experiment (MGEX) stations and simulated LEO signals considering the above-mentioned errors, the results demonstrated that the convergence time and the positioning accuracy are strongly degraded, sometimes even getting worse than the GNSS-only solutions in case of code biases at the level of tens of ns. These findings underscore the challenges to fully realize the potential of LEO-augmented PPP in real-world applications.
BibTeX
@inproceedings{ye2025impact,
  author = {Zihao Ye and Kan Wang and Chunbo Wei and Beixi Chen and Jinqian Wang and Meifang Wu},
  title = {{The impact of LEO satellite hardware delays on LEO-augmented precise point positioning}},
  booktitle = {GPS},
  year = {2025},
  doi = {10.1007/s10291-025-01995-6},
}
Figure