⚡ electrical engineering

Earth-baseline VLBI restores the observability of a lunar surface station in joint orbit-and-clock determination

This paper demonstrates that Earth-baseline Very Long Baseline Interferometry (VLBI) is essential for restoring the absolute observability of lunar surface stations in joint orbit-and-clock estimation, particularly when satellite constellation geometry is insufficient to resolve positional datum defects, thereby significantly improving positioning accuracy from tens of meters to under 10 meters.

Original authors: Chakshu Baweja

Published 2026-07-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Chakshu Baweja

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 build a map of a new city on the Moon. You have a team of surveyors (satellites) floating in the sky and a single, crucial landmark (a station) sitting on the ground. Your goal is to tell everyone exactly where that ground station is located relative to the Earth.

This paper asks a simple but critical question: Can the surveyors in the sky figure out the ground station's exact location just by talking to each other and to the station?

The authors say: No, not always. And here is why, explained through a few analogies.

The "Floating Island" Problem

Imagine your surveyors and the ground station are all standing on a giant, invisible raft floating in a foggy ocean. They can measure the distance between each other perfectly. They know, "I am 10 meters from Bob, and Bob is 5 meters from Alice."

However, because they are all on the same raft, they don't know where the raft is in the ocean. The whole raft could be drifting north, south, or spinning around, and their internal measurements wouldn't change. In math terms, the paper calls this a "datum defect." The group knows their shape, but not their place.

To fix this, they need a rope thrown from the shore (Earth) to tie the raft down.

Two Ways to Throw the Rope

The paper identifies two ways to tie this lunar raft to Earth, and they work very differently depending on how many surveyors you have.

1. The Indirect Rope (Earth → Satellites → Station)
This is like throwing a rope to the surveyors in the sky, and hoping they can pass a message down to the ground station.

How it works: Earth measures the distance to the satellites. The satellites then measure the distance to the ground station.
The Catch: This only works if the surveyors are spread out nicely in the sky. If you only have three surveyors clustered in one corner of the sky, they can't "see" the ground station from enough different angles to pin down its exact location. The "rope" slips, and the ground station remains lost in the fog.
The Result: If your satellite group is "sparse" (too few or poorly placed), this method fails completely. The ground station's location is mathematically "unobservable."

2. The Direct Rope (Earth → Station)
This is like throwing a rope directly to the ground station, bypassing the satellites entirely. The paper calls this VLBI (Very Long Baseline Interferometry).

How it works: Earth uses two or more radio telescopes to listen to a signal from the ground station. By comparing the tiny difference in when the signal arrives at each telescope, Earth can triangulate the station's exact spot.
The Magic: This works regardless of how many satellites you have. Even if you have only one satellite, or none at all, this direct rope ties the station to Earth.

The Big Discovery: "Restore" vs. "Sharpen"

The paper's main finding is a rule for mission planners: It depends on your satellite team.

Scenario A: The Sparse Team (The "Restore" Case)
If you have a small, early constellation (like just 3 satellites), the Indirect Rope fails. The ground station is lost.
- Adding the Direct Rope (VLBI): This doesn't just make things slightly better; it restores the ability to see the station at all. It pulls the station out of the "unobservable" zone and pins it down to within about 20 meters. Without this, the error is infinite.
- Analogy: It's the difference between having a map of a city that is completely blank versus having a map that shows the streets.
Scenario B: The Rich Team (The "Sharpen" Case)
If you have a large, well-spread-out constellation (like 6+ satellites), the Indirect Rope works fine on its own. The station is already observable.
- Adding the Direct Rope (VLBI): This doesn't fix a broken system; it just sharpens the picture. It improves the accuracy from about 23 meters down to 10 meters. It's a nice upgrade, but not a life-or-death necessity.

The "Three-Telescope" Rule

The paper also found a specific rule for how many Earth telescopes you need to make the Direct Rope work.
Because the Moon is far away, the angle between Earth telescopes looks very small from the Moon's perspective.

1 Telescope: Tells you the station is somewhere on a line.
2 Telescopes: Tells you the station is somewhere on a flat sheet (plane).
3 Telescopes (not in a straight line): Finally, you get the 3D location.
You need at least three non-collinear Earth stations to fully fix the station's position in a single snapshot.

The Bottom Line

The authors built a computer engine to prove these math rules. They found that:

Internal measurements alone (satellites talking to each other) can never fix the absolute location of a lunar base; they always leave a "floating" uncertainty.
VLBI (Direct Rope) is the only thing that can fix a sparse satellite network.
For big networks, VLBI is just a bonus that makes the map more precise.

The paper concludes that for future lunar missions (like ESA's "NovaMoon" station), if you are starting with a small satellite constellation, you must have a direct Earth-baseline link (like a VLBI transmitter) to know where you are. If you have a huge constellation, you can skip it, but it's still helpful to have.

Note: The paper emphasizes that these results are based on computer simulations of a "representative" lunar network, not on real data from a live mission yet. However, the mathematical rules governing the geometry are proven to be exact.

Technical Summary: Earth-baseline VLBI restores the observability of a lunar surface station in joint orbit-and-clock determination

Problem Statement
As lunar positioning, navigation, and timing (PNT) transitions from concept to hardware (e.g., ESA's Moonlight/LCNS, NovaMoon, and LunaNet), a fundamental estimation challenge remains: determining the absolute orbits and clocks of lunar infrastructure and tying them to an Earth or inertial frame. While internal lunar measurements (station-to-satellite and inter-satellite ranging) constrain the relative geometry of a lunar network, they leave a six-dimensional rigid-body datum defect (three translations and three rotations). The core question addressed is which measurements render the absolute position of a lunar surface station observable, and under what conditions a tie to the Earth/inertial frame is strictly necessary versus merely beneficial.

Methodology
The study employs a static, single-epoch snapshot batch least-squares estimator (Gauss–Newton) to analyze the joint determination of a lunar surface station's position, a constellation of lunar satellites, and their respective clocks.

Observables: The model incorporates four measurement types: Earth-baseline geodetic Very Long Baseline Interferometry (VLBI) delay to the lunar beacon, lunar-local ranging (station-to-satellite and inter-satellite), Earth-to-satellite ranging, and a station clock-sync pseudo-observable.
Regimes: The analysis evaluates two distinct constellation regimes: a "sparse" regime (3 satellites) where the constellation geometry is insufficient to propagate an Earth anchor, and a "rich" regime (6 satellites) where the geometry is robust.
Validation Framework: The study utilizes the kshana open-source engine. The Fisher Information Matrix (FIM) and Cramér–Rao Lower Bound (CRLB) engine is explicitly Validated against independent libraries (NumPy) and closed-form bounds (Kay). However, the specific lunar network results are labeled Modelled, as they rely on a representative simulation rather than real flight tracking data.
Controls: The analysis distinguishes between information gain and observation count by varying VLBI noise levels while holding observation counts constant. It also employs a 200-seed ensemble to ensure robustness against specific truth/noise draws.

Key Contributions

Theorem 1 (Internal Datum Defect): The paper proves that for a non-degenerate lunar cluster, internal ranging measurements alone leave a six-dimensional null space in the Fisher Information Matrix corresponding to rigid-body translations and rotations. With clock synchronization, the residual defect is purely positional.
Conditional Design Law: The paper establishes that an Earth-baseline tie is not universally "helpful" but conditionally necessary.
- Indirect Tie (Earth $\to$ Satellite $\to$ Station): This propagates the Earth anchor through the constellation. It only succeeds if the constellation geometry is "rich" enough to propagate the datum to the station.
- Direct Tie (Earth-baseline VLBI): This ties the station beacon directly to the Earth frame, independent of the satellite constellation.
- Verdict: In sparse constellations, the indirect tie fails, leaving the station's absolute position in the null space (unobservable). VLBI is required to restore observability. In rich constellations, the indirect tie already suffices, and VLBI merely sharpens the error bound.
Three-Station Threshold: The paper derives a geometric threshold for the direct VLBI tie. Because a single VLBI baseline at a single epoch is rank-one (constraining only two axes in the plane of the sky), the absolute position becomes observable only when three non-collinear Earth stations are available. Below this threshold, the datum defect remains non-zero.

Results

Sparse Regime (3 Satellites): Without VLBI, the station's absolute position is unobservable (infinite error bound). Adding VLBI restores observability. In a 200-seed ensemble, the median station error improves by a factor of 91× (from a median of ~793 m to ~13.4 m). The Cramér–Rao bound for the absolute position is 20.1 m with VLBI.
Rich Regime (6 Satellites): The indirect tie alone renders the station observable with a CRLB of 23.2 m. Adding VLBI tightens this bound to 9.7 m (a ~2.4× improvement), demonstrating a "sharpening" rather than "restoration" effect.
Estimator Efficiency: The Gauss–Newton estimator achieves the theoretical CRLB with an RMS-to-bound efficiency of 1.02 and a bias under 4% of the RMS, confirming that the reported limits are information-theoretic, not solver artifacts.
Orbital Robustness: The results hold for representative elliptical lunar frozen orbits (ELFO) used in Moonlight/LCNS designs, confirming the design law is a property of the estimation geometry rather than specific circular orbit assumptions.

Significance and Claims
The paper reframes the Earth-baseline VLBI tie from a generic enhancement to a conditional architectural requirement.

Design Implication: Mission planners must assess whether their constellation geometry is "rich" enough to carry the absolute datum. If the constellation is sparse, early, or geometrically constrained (e.g., polar), an Earth-baseline VLBI tie (such as a transmitter on a NovaMoon-class station) is mandatory to achieve any finite error bound on the station's absolute position. If the constellation is rich, VLBI is optional but improves precision.
Honesty in Modeling: The authors explicitly distinguish between the Validated nature of their estimation engine (mathematically proven against independent libraries) and the Modelled nature of the lunar network results (based on simulation). They do not claim these numbers represent flight heritage or real-world tracking performance, but rather the theoretical limits of the proposed geometry.
Reproducibility: The work is fully open-source, providing a toolkit that allows mission teams to regenerate observability figures for specific scenarios, varying constellation sizes, station counts, and design criteria (e.g., E-optimality for safety-of-life applications).

The paper concludes that the "three-station threshold" and the "restore-versus-sharpen" design law are exact algebraic properties of the estimation model, providing a rigorous basis for sizing lunar PNT infrastructure before real tracking data becomes available.

The "Floating Island" Problem

Two Ways to Throw the Rope

The Big Discovery: "Restore" vs. "Sharpen"

The "Three-Telescope" Rule

The Bottom Line

More like this