Sovereignty in Quantum Navigation Starts at Acceptance Testing

In April 2024, Finnair stopped flying to Tartu. The airport’s only instrument approach relies on a satellite signal, and persistent GPS interference over the Baltic had made that approach unusable in poor visibility. Finnair said in its suspension announcement at the end of April 2024 that two of its aircraft had turned back mid-approach in the weeks before, and it resumed service about a month later, once an alternative approach procedure that did not depend on the satellite signal was available.

Nothing about that is exotic. GPS is the United States constellation within the wider Global Navigation Satellite System (GNSS) family, and its civil signal arrives at the ground below the background radio noise, recovered only because the receiver knows exactly what pattern to look for. It can be swamped by a low-power transmitter of the kind sold openly online. Satellite navigation is a shared civil and military utility with a single point of failure, and everyone who depends on it has known this for twenty years.

Quantum sensing is the first credible answer that does not involve launching more satellites. Most of the sovereignty debate around quantum technology has been about cryptography and about machines that do not exist yet. Positioning and timing hardware is different: it’s in flight trials now, and the procurement decisions are arriving before most buying organisations have anyone who can assess them. This article sets out what a buyer has to be able to do for a sovereignty claim about quantum navigation to mean anything.

What the sensors measure

Positioning, navigation, and timing (PNT) is the layer underneath almost everything else. It tells a vehicle where it is, tells a network when an event happened, and tells a weapon where to go. Satellite constellations supply all three cheaply, which is why they are the target.

The fallback is an inertial navigation system (INS): accelerometers and gyroscopes that measure motion directly, with no external signal at all. The catch is arithmetic. Position comes from integrating acceleration twice, so every small bias in the sensor grows without limit as the mission runs. That growth is called drift, and it is the whole game. A navigation-grade inertial system drifts unaided by anything from a fraction of a nautical mile to several nautical miles per hour, depending on the grade of the sensors, and the figure that matters to a buyer is the one for the class they can afford to fly. For a submarine on a long patrol or an aircraft crossing a jammed region, that’s the number that decides whether the mission works.

Cold-atom sensors attack the drift directly. An atom interferometer cools a small cloud of atoms with lasers until they are almost stationary, then uses pulses of light to split the cloud along two paths and bring it back together. Acceleration or rotation along the way changes the phase between the two paths, and the resulting interference pattern reads out the motion. The important property for a buyer is not the sensitivity headline. It’s that the scale factor depends on the wavelength of a laser and on atomic constants rather than on a machined part that ages, so the sensor has far less to recalibrate over its life.

Three other device families follow from the same physics. A quantum gravimeter measures local gravity by dropping cooled atoms and timing the fall against the same interferometer trick; a gradiometer runs two of them at a known separation and subtracts, which cancels the vibration of the platform they sit on and leaves the underground density anomaly you were looking for. Magnetic anomaly navigation uses a very sensitive magnetometer to read the fixed magnetic pattern of the Earth’s crust and match it against a stored map. Optical clocks replace the microwave transition used in caesium standards with an optical one running on the order of fifty thousand times faster, and laboratory versions now reach fractional uncertainties near one part in 10 to the 18th.

None of this removes drift. It slows it, and it does so by trading against size, weight, power, and warm-up time. Every specification a vendor shows you is a point on that trade.

Demonstrated, in trials, and still in the laboratory

We tell the teams we train to sort every quantum sensing claim into one of three bins before reading anything else, because the bins have very different procurement consequences.

Fielded. Atomic timing is the mature end. Caesium and rubidium frequency standards have been in service for decades, chip-scale atomic clocks are catalogue items, and holdover timing for a network under GNSS loss is an engineering problem with known costs. Gravity surveying is close behind. In February 2022, a University of Birmingham team led by Michael Holynski published the first outdoor detection of a buried tunnel using a quantum gravity gradiometer, in Nature (Stray et al., volume 602, pages 590-594), working through the traffic vibration and thermal noise that had kept the technique indoors.

In trials. Quantum inertial navigation sits here. Imperial College London and M Squared demonstrated a transportable quantum accelerometer, shown publicly at the UK National Quantum Technologies Showcase in November 2018, and in May 2024 the UK Ministry of Defence announced commercial airborne trials of a quantum inertial navigation payload, flown with Imperial College London and QinetiQ. Magnetic anomaly navigation is at a similar stage, with several vendors publishing flight and ground trial results against inertial-only baselines. Trial results are real evidence, and they are also selected. The comparison baseline, the duration, and the terrain are all choices the vendor made.

Laboratory. Quantum radar belongs here, and the market talk around it has run well ahead of the physics. The technique, quantum illumination, sends one photon of an entangled pair at a target and keeps its twin as a reference, which in principle picks out a faint return from heavy noise. The demonstrated microwave experiments work over short distances inside cryostats, and the entanglement that supplies the advantage is destroyed by exactly the loss and thermal noise that a real atmospheric path imposes. It’s a live research question. It’s not a procurement category, and any proposal that treats it as one is either misjudging the physics or selling ahead of the evidence.

Two further habits are worth building into the same reflex. Distinguish announced from demonstrated, and demonstrated from repeated by someone else. And keep physical and logical performance straight in your head: a sensitivity figure achieved in a shielded basement with a two-hour warm-up is not the figure you will get on a vehicle.

Autonomy, not autarky

The reflex definition of sovereign capability is domestic manufacture of every part. For quantum sensing that definition is unusable. The supply chain runs through a small number of suppliers for stabilised lasers, ultra-high-vacuum components, isotopically enriched sources, photodetectors, and magnetic shielding, and almost no country covers all of them. A national programme built on the autarky definition either fails or produces something nobody deploys.

The workable definition is narrower and harder to fake. Sovereign capability is the ability to verify, integrate, harden, and sustain a system without depending on the party that sold it to you. You can buy the box. What you can’t buy from the same vendor is an independent answer to the question of whether the box does what the datasheet says under the conditions you will actually use it in.

This is also where the dual-use argument earns its keep, and it earns it in a specific way rather than a general one. Defence money pays for the first units, which are expensive and awkward. Civil markets then widen the supplier base that defence later draws on. Optically pumped magnetometers developed for detecting buried and concealed objects now sit inside wearable brain scanners, and the companies building those scanners keep a magnetometer supply chain alive that no single defence programme could sustain on its own. A country with a live commercial sensor sector has second sources during a crisis. A country with one classified programme has a single point of failure with a security clearance.

Six assurance questions

These are the questions we work through with procurement and assessment teams. Each one has a reason attached, and the reason is the part that survives when the technology changes.

Drift, and the conditions attached to it

Ask for the drift figure, then ask for every condition under which it was obtained. Ambient temperature range. Vibration spectrum. Warm-up time from cold. Duration of the run. Whether the system was aided by any external input during the measurement, including a GNSS fix at the start. An unaided quantum INS quoted at a drift rate that assumed a satellite fix every hour is not an unaided system, and the distinction disappears completely on the summary slide. Require the raw comparison against whatever inertial system you field today, over the mission profile you actually fly.

Calibration you can perform yourself

A sensor you can’t check is a sensor you’re trusting. Establish before contract award whether your own metrology institute or laboratory can calibrate the device against a traceable reference, what equipment that requires, and what it costs per unit per year. Cold-atom devices are attractive here precisely because much of their calibration is set by physical constants, and that argument only helps you if you can verify it locally. If the answer is that the unit ships back to the manufacturer annually, you’ve bought a standing dependency on a foreign laboratory and should price it that way.

The reference map is part of the system

This is the dependency most buyers miss. Magnetic anomaly navigation and gravity-matching navigation both work by comparing a live measurement against a stored map of the Earth. The sensor is hardware you can inspect. The map is a dataset, it has an owner, it has an update cycle, and it may have been produced by a foreign government or a commercial survey firm under licence. If the licence lapses or the resolution over your area of interest is poor, your jam-resistant navigation system stops working, and no amount of hardware sovereignty helps. Treat map provenance, licence terms, update rights, and the ability to survey your own territory as line items in the specification.

Adversarial testing

Quantum sensors are immune to the specific attack of GPS jamming. They are not immune to attack. A magnetometer-based system can be fed a false local field. An atom interferometer can be disturbed by vibration and by stray magnetic fields, and the failure may be silent rather than obvious. A hybrid system that fuses quantum, classical inertial, and satellite inputs can be attacked through whichever input the fusion algorithm currently weights most heavily. Acceptance testing should include a red team whose job is to make the system report a confident wrong answer, and the pass criterion should include how the system behaves when it is being fooled. A sensor that degrades loudly is worth more than one that degrades quietly.

Components, spares, and second sources

Map the bill of materials down to the parts with one supplier. For most of these devices that list is short and predictable: the seed laser, the vacuum assembly, specific photodetectors, magnetic shielding alloy. For each one, record the country of origin and whether it falls under an export control regime that a future government could tighten. Then decide, item by item, whether you want a stockpile, a qualified second source, a domestic substitute at reduced performance, or an accepted risk. Contracts should carry spares provisioning and a support horizon that matches the platform life, not the vendor’s product roadmap.

Interfaces, sustainment, and the exit

Require documented, open interfaces for data out and for configuration in, and confirm the unit runs with no connection to a vendor cloud service. Remote calibration updates are convenient and they are also a control channel and a supply chain risk. Fund a training package and technical documentation at a level that lets your own technicians replace a laser module and re-verify the unit. And write the exit before you sign it: source code and design escrow, modular construction so a component can be replaced with an equivalent from another supplier, and a defined second path if the vendor relationship ends. Sovereignty here is measured by what you can still do on the day the supplier says no.

Who asks the questions

Every question above sits between two professions that rarely overlap. Answering the drift question needs someone who understands what an interferometer measures and why the warm-up time is what it is. Turning that understanding into a contract clause needs someone who knows how acceptance criteria, escrow, and support horizons actually work. Programmes fail this test in both directions. A physicist evaluates the hardware beautifully and signs a contract with no exit. A commercial lead writes an excellent contract against a specification that quietly assumes a satellite fix.

The competence is teachable, and it is narrower than a physics degree. What a technical assessor needs is enough of the underlying mechanism to read a datasheet critically, a working map of what has been demonstrated against what has been announced, and a structured method for turning that into testable acceptance criteria. We build our sensing material around that combination, because the alternative is a procurement organisation that can only repeat what vendors tell it.

The UK made sovereign PNT one of the named priorities of its 2023 national quantum strategy, and other governments have followed with similar language. The strategies are funded. The specification and acceptance capability, in most cases, is not.

Where to start

If your organisation will be evaluating quantum PNT or sensing systems in the next few years, the useful first move is to run the six questions above against whatever proposal is nearest to hand, and see which ones your team cannot currently answer. That gap is your training requirement, and it’s usually concrete.

Quantum Academy’s sensing and PNT material is written for the people who have to specify, test, and accept these systems rather than build them, and you can find the current programmes at quantumacademy.com/. For the wider strategic argument behind this piece, including the industrial policy dimension we have only touched on here, see the original analysis on PostQuantum.com. Anyone weighing up the assessor role itself as a direction of travel will find the relevant paths mapped at QuantumCareers.com.