🔭 astrophysics

Ephemeris and Almanac Design for Lunar Navigation Satellites

This paper proposes a hybrid Chebyshev-Fourier message representation for lunar navigation satellites that achieves sub-meter positioning accuracy and compact data sizes within the LunaNet specification, effectively supporting both precise ephemeris generation and reliable almanac-based visibility identification across various lunar orbit regimes.

Original authors: Keidai Iiyama, Grace Gao

Published 2026-02-18

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Keidai Iiyama, Grace Gao

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to navigate a car in a city where the roads are constantly shifting, the gravity is weird, and you only have a tiny, 900-character text message to tell your GPS where you are. That is essentially the challenge of navigating on the Moon.

This paper, written by researchers at Stanford, proposes a new "language" for sending navigation data to the Moon. Here is the breakdown in simple terms:

1. The Problem: The Moon is a Bumpy, Unpredictable Ride

On Earth, GPS satellites fly in nice, predictable circles. You can describe their path with a simple math formula.

But on the Moon, it's chaotic:

The Gravity is Lumpy: The Moon isn't a perfect sphere; it has "mass concentrations" (heavy spots) that pull satellites off course.
The Earth is a Bully: The Earth's gravity tugs on lunar satellites, especially when they are far away from the Moon.
The Data Limit: The Moon's new internet (called LunaNet) has a very strict data limit. You can only send about 900 bits (tiny chunks of data) per message. If you try to send a full, detailed map of the satellite's path, it won't fit.

The Analogy: Imagine you are trying to describe the path of a rollercoaster that is being pushed by a giant magnet (Earth) and has a bumpy track (Moon's gravity). You have to write this description on a postcard (the 900-bit limit). If you try to write every twist and turn in detail, the postcard will be too heavy to mail.

2. The Solution: A "Hybrid" Navigation Message

The authors created a new way to compress this complex path into a tiny message. They combined three different mathematical tools, like mixing ingredients in a recipe:

Osculating Elements (The Skeleton): This is the basic "shape" of the orbit. It's like knowing the rollercoaster is generally a big loop.
Chebyshev Polynomials (The Smooth Curve): These are math tools that are great at fitting smooth curves. They handle the general, slow changes in the satellite's path.
Fourier Series (The Rhythm): This captures the "wobbles" and repeating patterns caused by the Moon's rotation and the Earth's pull.

The Analogy:
Think of the satellite's path as a song.

The Osculating Elements are the main melody (the tune you hum).
The Chebyshev Polynomials are the smooth, slow background notes.
The Fourier Series are the specific drum beats or high-pitched riffs that repeat every time the satellite goes around the Moon.

By combining the melody, the background, and the rhythm, they can describe the entire song using very few notes. This allows them to fit a highly accurate 6-hour (or even longer) path into that tiny 900-bit postcard.

3. The Two Types of Messages

The paper designs two specific types of messages for two different needs:

A. The Ephemeris (The "Live Map")

What it is: A high-precision, short-term map.
Goal: To tell a user exactly where the satellite is right now and where it will be in the next few hours.
Performance: The new method is so good that it can predict the satellite's position within less than a meter (about the length of a ruler) and its speed within less than a millimeter per second.
Why it matters: This is needed for landing a rover or guiding a human to a specific rock on the Moon.

B. The Almanac (The "Long-Term Schedule")

What it is: A low-precision, long-term schedule.
Goal: To help a receiver turn on and quickly figure out, "Which satellites are visible right now?" without needing a super-precise location.
Performance: It predicts the satellite's location over 15 days (covering a full lunar night). While it's not precise enough to land a rover, it's accurate enough to say, "Look up at 2 o'clock, the satellite will be there."
Why it matters: If your GPS receiver has been off for a week (like during the long lunar night), it needs this "Almanac" to wake up and know where to look for signals immediately.

4. The Results: It Works!

The researchers tested this on four different types of lunar orbits (some high and elliptical, some low and circular).

The Verdict: Their new "Hybrid Recipe" fits the data perfectly within the 900-bit limit.
The Win: They managed to keep the messages small (like a tweet) while keeping the accuracy incredibly high (like a laser pointer).

Summary

This paper is about inventing a new, super-efficient way to talk to the Moon. By mixing different math techniques, the authors created a system that can send highly accurate navigation data over a very narrow "bandwidth" pipe. This is a crucial step toward making the Moon a place where we can drive cars, land rovers, and explore safely, just like we do on Earth today.

1. Problem Statement

The rapid acceleration of lunar exploration (2024–2030) necessitates robust Positioning, Navigation, and Timing (PNT) services via the LunaNet architecture. A critical component is the Lunar Augmented Navigation Service (LANS), a constellation of navigation satellites. However, designing navigation messages (ephemeris and almanac) for lunar satellites presents unique challenges distinct from terrestrial GNSS:

Complex Dynamics: Lunar satellites operate in highly elliptical orbits (e.g., Elliptical Lunar Frozen Orbits - ELFOs) subject to strong third-body perturbations (Earth/Sun) near apolune and significant lunar gravity anomalies near perilune. Simple Keplerian elements with minor corrections are insufficient.
Data Constraints: The LunaNet Interoperability Specification limits the ephemeris message size to approximately 900 bits (after accounting for clock and health data).
Accuracy Requirements: Strict thresholds are required: 40 m position and 10 mm/s velocity (95th percentile), with LCRNS requiring even tighter bounds (13.43 m, 1.2 mm/s).
Operational Gaps: Previous studies focused on short-duration ephemeris fitting (typically <4 hours) and largely ignored almanac design, which is essential for cold/warm-start receiver operations (predicting satellite visibility over long periods).

2. Methodology

The authors propose a unified framework for generating compact, high-accuracy ephemeris and almanac messages using a hybrid mathematical approach.

A. Coordinate Frames and Dynamics

Frames: Utilizes Moon-Centered Inertial (MCI), Moon Principal Axis (PA), and Principal Axis Inertial (PAI) frames.
Propagation: Uses a high-fidelity dynamics model (Moon gravity up to degree/order 80 + Earth/Sun third-body perturbations) to generate reference trajectories.

B. Ephemeris Representation (Short-to-Medium Term)

The proposed ephemeris model combines three components to fit satellite trajectories over arcs up to 480 minutes:

Osculating Orbital Elements (OE): Five elements ( $a, e, i, \lambda, \omega$ ) define the nominal two-body motion.
Chebyshev Polynomials: Used to model the residuals of the Cartesian position and velocity. Chebyshev polynomials minimize the maximum approximation error (minimax property), allowing high accuracy with fewer coefficients.
Fourier Series: Added to capture specific periodic variations (primarily those related to the orbital period) that Chebyshev polynomials alone might miss over longer arcs.

Fitting Strategy: Uses least-squares fitting with Chebyshev-Lobatto nodes to avoid Runge's phenomenon.
Bit Allocation: Parameters are quantized based on sensitivity analysis to ensure the total message size remains under 900 bits while meeting error tolerances.

C. Almanac Representation (Long-Term)

Designed for 15-day validity to support warm starts after a lunar night:

Model: A compact low-order polynomial + single-term Fourier series applied directly to osculating orbital elements.
Frequency Selection:
- Semi-major axis ( $a$ ): Modeled with the orbital period ( $T_{orb}$ ).
- Other elements ( $e, i, \Omega, M, u$ ): Modeled with the lunar sidereal period ( $T_{sid} \approx 27.3$ days) to capture third-body and gravitational coupling effects.
Goal: Provide coarse geometry sufficient for visibility prediction and Doppler search space reduction, rather than high-precision positioning.

3. Key Contributions

Hybrid Ephemeris Formulation: Introduces a novel OE + Chebyshev + Fourier combination. This outperforms previous methods (OE+Chebyshev or Chebyshev-only) by capturing both secular trends and periodic perturbations more efficiently within the 900-bit limit.
First Lunar Almanac Design: Proposes the first dedicated almanac model for lunar navigation satellites, enabling reliable satellite visibility identification for warm-start operations over 15-day arcs.
Comprehensive Validation: Validates the framework across four distinct orbit regimes:
- 30-hour ELFO (LCRNS)
- 24-hour ELFO (LCNS)
- 12-hour ELFO (LNSS)
- 6-hour Polar Orbit
Optimized Bit Allocation: Develops a systematic method to determine bit resolution for parameters to strictly adhere to the 900-bit constraint without sacrificing accuracy.

4. Key Results

Ephemeris Performance (900-bit Budget)

Accuracy: The hybrid model achieves sub-meter position accuracy (e.g., 0.32 m for 30-hr ELFO) and sub-millimeter-per-second velocity accuracy (e.g., 0.70 mm/s) for fitting arcs up to 240 minutes (4 hours).
Efficiency: For arcs $\ge$ 240 minutes, the addition of Fourier terms significantly reduces errors compared to OE+Chebyshev alone, allowing lower-order Chebyshev polynomials and saving bits.
Orbit Regimes:
- ELFOs: Achieved <1 m position error for 30h and 24h orbits; <2.5 m for 12h orbits within 900 bits.
- Polar Orbit: Achieved <0.21 m position error for 6-hour arcs.
Comparison: The proposed method meets the stringent LCRNS SISE requirements (13.43 m / 1.2 mm/s) with a large margin, whereas prior methods often struggled to extend validity beyond 2–4 hours without exceeding bit limits.

Almanac Performance (15-Day Arc)

Visibility Prediction: Successfully predicts satellite visibility from the lunar South Pole with >97% accuracy (99.4% for 30-hr ELFO).
Angular Errors: Median elevation and azimuth errors are generally < 2°.
Search Space Reduction:
- Range Error: < 250 km (99th percentile) for ELFOs; < 80 km for Polar orbits.
- Doppler Error: < 105 Hz (99th percentile) for ELFOs; < 2.2 Hz for Polar orbits.
These errors are small relative to the true signal ranges and Doppler shifts, effectively narrowing the acquisition search space for user receivers.

5. Significance

This paper provides a critical foundation for the operational implementation of LunaNet. By solving the trade-off between accuracy, validity duration, and message compactness, the proposed framework enables:

Extended Ephemeris Validity: Reducing the frequency of orbit updates required from ground tracking assets, which are scarce in the lunar environment.
Robust Receiver Operations: Enabling receivers to perform warm starts quickly by accurately predicting satellite visibility and signal parameters after long periods of data loss (e.g., during the lunar night).
Interoperability: The design adheres to the LunaNet Interoperability Specification, ensuring that navigation messages from different agencies (NASA, ESA, JAXA) can be processed by common user equipment.

The work bridges the gap between high-fidelity orbital dynamics and the practical constraints of broadcast navigation messages, paving the way for reliable autonomous navigation in cislunar and lunar space.