Research on high-precision time synchronisation technology for sea mobile platforms

Dao Peng Dong, Hong Shuo Wu, Qing Guo, Jin Yang, Xi LiInternational Journal of Information and Communication Technology 2024 · DOI

TL;DR

This paper presents a velocity-aided smoothing algorithm designed to achieve sub-10 nanosecond time synchronization for mobile platforms at sea. By leveraging high-precision velocity data from GNSS carrier loops to refine real-time positioning, the researchers successfully minimized the timing errors that typically plague dynamic "Common View" satellite synchronization.

What Problem Does This Paper Address?

In modern naval operations, "spatiotemporal unity" is critical. For a fleet of ships to coordinate defensive actions or share sensor data effectively, every platform must operate on a unified, high-precision clock.

While static stations can achieve timing accuracy within 3–5 nanoseconds using GNSS Common View (CV)—a technique where two stations observe the same satellite simultaneously to cancel out errors—mobile platforms struggle. On a moving ship, position errors translate directly into timing errors. Because light (and radio signals) travels at roughly 30 cm per nanosecond, a mere 3-meter error in a ship's calculated position introduces a ~10 ns error in time synchronization. In the unpredictable environment of the open sea, maintaining the decimeter-level positioning required for nanosecond-scale timing is a significant hurdle.

Methods

The authors propose a position’s smooth filtering with velocity algorithm to bridge the gap between static and dynamic performance.

  1. Velocity-Aided Smoothing: The core innovation relies on the fact that a GNSS receiver's carrier loop can measure velocity with an order of magnitude more accuracy than it can measure raw position. The researchers implemented an $\alpha$-$\beta$ filter (a simplified state-space observer) that uses this high-precision velocity to predict the platform's next position. This prediction is then balanced against the latest raw position measurement to produce a "smoothed" estimate.
  2. Error Budgeting: The paper meticulously accounts for various environmental errors, including:
  3. Ionospheric/Tropospheric Delays: Using dual-frequency correction models and the "Hopfield" model.
  4. Satellite Ephemeris Errors: Calculating how orbital uncertainties impact stations separated by up to 2,000 km.
  5. Hardware Optimization: To prevent internal electronics from introducing jitter, the team used an FPGA-based signal processing module. They implemented a unified global reset and a synchronous clock enable scheme to ensure that signal processing delays across different channels remain perfectly consistent.

Datasets

The researchers evaluated their approach through three distinct scenarios:

  • Dynamic Simulator: A one-hour simulated trajectory mimicking a ship's sinusoidal oscillation in three axes (X, Y, and Z) caused by wind and waves.
  • Rooftop Mobile Test: A real-world scenario where a GNSS antenna was manually moved horizontally to test the filter's response to sudden motion.
  • Lake Surface Field Test: Conducted at the Tiantai Mountain Reservoir in Chengdu, China. This involved two ships equipped with antennas in a short-baseline configuration to verify the CV results in a representative marine environment.

Key Insights and Ideas

  • The "Velocity is King" Realization: The most important insight is that raw GNSS position data is too "noisy" for nanosecond timing on a moving platform. By treating velocity as the primary reliable metric and position as a secondary correction, they achieved a positioning accuracy of approximately 1 meter in dynamic conditions.
  • Hardware-Software Synergy: The paper emphasizes that high-level math isn't enough. Without FPGA-level control over internal hardware delays (ensuring "zero clock delay" to each module), the algorithmic gains would be lost to electronic noise.
  • The 10ns Threshold: The authors demonstrate that even after accounting for multipath effects, device noise, and atmospheric residuals, the cumulative error budget ($\sigma_{dev}$) stays around 5 ns, comfortably meeting the 10 ns requirement for sea mobile platforms.

Key Results

The proposed method showed consistent high performance across all tests:

  • Theoretical Estimate: The statistical sum of all error sources was calculated at 5.013 ns.
  • Simulator Performance: In the sinusoidal motion test, the timing standard deviation was 2.87 ns.
  • Rooftop Test: Manual movement resulted in a standard deviation of 3 ns, with no significant "jumps" in time difference during motion.
  • Lake Test: The most realistic trial on the Tiantai Mountain Reservoir yielded a standard deviation of 2.19 ns.

These results confirm that the system can maintain a synchronization accuracy of 10 ns or better, even when the platforms are in motion.

Limitations and Future Work

While the results are promising, the paper focuses primarily on short-baseline tests (where the two receivers are relatively close). Although they provide a theoretical error budget for stations 2,000 km apart, they did not present experimental data for long-distance dynamic synchronization.

Additionally, the authors mention that a rubidium atomic clock can be integrated into the device to maintain timing during GNSS signal interruptions (holdover). However, the specific performance of this "hybrid" system—how long it can maintain <10 ns accuracy without a satellite fix while the ship is tossing at sea—remains a topic for future detailed study.

Summary updated

Abstract

Under static conditions, at present, mature technologies for long-distance and high-precision time synchronisation include satellite common view (CV) and two-way satellite time and frequency transfer (TWSTFT) and so on. However, under dynamic conditions, research on high-precision time synchronisation technology is relatively lacking such as mobile platforms on the sea. Considering the dynamic conditions of mobile platforms on the sea, a method of position's smooth filtering with velocity is proposed to improve the accuracy of position measurement, reducing the time measurement error introduced by position error. The proposed method ultimately improves the CV comparison accuracy between sea surface mobile platforms. The simulation and actual test results show that by using the method of position's smooth filtering with velocity, the CV comparison accuracy between mobile platforms on the sea can reach 10 ns.
BibTeX
@article{dong2024research,
  author = {Dao Peng Dong and Hong Shuo Wu and Qing Guo and Jin Yang and Xi Li},
  title = {{Research on high-precision time synchronisation technology for sea mobile platforms}},
  journal = {International Journal of Information and Communication Technology},
  year = {2024},
  doi = {10.1504/ijict.2024.138561},
}
Figure