Redundancy, Training and Old Physics: What Carries Logistics Through a Satellite Outage

Future Insight — part of the Xin.bz Future Insights series.

At a glance

• 14 to 40 days — holdover from a cesium reference under ITU-T ePRTC specification, against 24 hours to a few days from an OCXO
• 89% — the position-accuracy improvement magnetic navigation delivered over conventional backup across four hours without GPS
• 50 metres — accuracy from Draper’s sliced-lens star tracker in GNSS-denied conditions, against 3 to 5 metres from a civil GPS receiver
• 20 metres — accuracy from South Korea’s operating eLoran chain, enough to hold a ship inside a dredged channel
• 6 sites — the United Kingdom’s Chain Odyssey eLoran network
• May 2026 — the month Britain opened its two-year Urgent Compass programme for miniaturised eLoran receivers and deployable stations
• Under one hour — the install time for magnetic navigation software on an airframe, software only
• $1 billion a day — the modelled cost of a 30-day GPS outage that all of this is measured against

Four hours without GPS

In 2026 an Embraer 170 left Puget Sound, turned north for southern Alaska, and flew the route with its GPS switched off. It navigated by the rock underneath it. Earth’s crust carries iron in patterns that vary from place to place and stay where they are, and a quantum magnetometer aboard the aircraft read that pattern against a survey map and returned a position. The flight ran four hours and twenty-three minutes. Position accuracy came out 89% better than the conventional backup.

A compass has read that same field for a thousand years. The new part is a sensor fine enough to tell one patch of seabed from another, and a map to match it against. Jamming the field means changing the geology.

That flight is the shape of the whole answer to contested orbits: old physics, new sensors, and a failure mode an attacker in orbit or on a hilltop leaves untouched.

The principle that organises everything

The companion piece set out the exposure: three satellite layers, each with a documented failure, and one attack in Iran that reached two of them through a single mechanism.

A backup earns the name by failing differently. That is the whole design principle, and most resilience spending misses it. A second GNSS receiver fails to the same jammer. A satellite phone held against a downed VSAT link fails to the same spoofed clock. Buying more of the same physics buys availability against equipment failure and nothing against interference.

Resilience comes from diversity of physics. Each alternative below reads a different property of the world, so an attack that defeats one leaves the others reporting.

Reference	Physics it reads	Defeated by
GNSS	Timing signals from orbit	Jamming, spoofing, orbital attack
eLoran	Ground-based low-frequency radio	Transmitter destruction
Inertial	Onboard acceleration and rotation	Time, through accumulated drift
Magnetic	Earth’s crustal magnetic field	Local anomalies and survey gaps
Celestial	Photons from stars	Cloud at visible wavelengths
Atomic holdover	An oscillator’s own stability	Time, through accumulated drift

Nothing in that right-hand column repeats. That is the point.

Redundancy: the receiver, the clock, the ground

Redundancy for logistics runs in three tiers, and the cheapest one comes first.

Tier one is the receiver. Multi-constellation, multi-frequency equipment tracks GPS, Galileo, GLONASS and BeiDou together across separate bands, and a jammer usually reaches one band. Running one receiver per constellation and comparing outputs exposes a spoof, because a spoofer that produces a consistent lie across four constellations and several frequencies is running a far harder attack. Interference detection in the receiver chipset turns silent failure into a flagged alarm, which is the difference between a crew that switches methods and a crew steering on a false position.

Tier two is the clock. This is where most operators discover their actual exposure, because timing outlasts position as a problem. The specification of the oscillator in a timing rack decides whether a GNSS denial is an inconvenience or a stoppage.

Reference	Drift	Holdover at 8 µs
OCXO	1–5 µs/hour	8 to 24 hours, to a few days
Rubidium	~36 ns/hour	Up to 7 days
Cesium, ITU-T ePRTC	Order of nanoseconds	14 to 40 days

Read that table as a planning horizon rather than a component list. A port, an exchange or a network with an OCXO survives a weekend. The same site with a rubidium reference survives a week, and with cesium it survives the kind of regional denial that has already happened in the Gulf. The upgrade is a capital line item measured in thousands, set against an outage cost measured in millions.

Tier three is the ground. A terrestrial signal breaks the dependency on orbit entirely, which is why the systems shut down in the GPS era are coming back.

New old physics: three references rebuilt

The striking pattern in current work is that the alternatives are old ideas returning with modern sensors and processing behind them.

Original	Modern form	Status
Loran-C ground radio, shut 2010	eLoran with differential corrections and miniaturised receivers	Operating in South Korea; six-site network in Britain
Magnetic compass, a millennium old	MagNav, quantum magnetometers reading crustal anomaly maps	Flown 4h 23m without GPS; drone install in under an hour
Sextant and star sight	Automated star tracker coupled to an inertial platform	50-metre accuracy demonstrated; patents filed
Celestial navigation by starlight	X-ray pulsar navigation using millisecond pulsars	Demonstrated autonomously in orbit, 2017

eLoran: the chain rebuilt

Loran-C was shut down in 2010 on the reasoning that GPS had settled the question. The rebuild treats low-frequency ground radio as the complement it always was. The signal arrives thousands of times stronger at the receiver than a satellite’s, propagates over the horizon, and resists spoofing, because faking it takes a transmitter the size of a building.

Britain moved furthest. The Chain Odyssey programme establishes a sovereign terrestrial backup across six transmission sites: Northern Ireland, the Shetlands, the Outer Hebrides, East Anglia and Dover, and southwest Cornwall. In May 2026 the Ministry of Defence opened Urgent Compass, a two-year programme for miniaturised receivers, antennas and transportable stations. South Korea operates three stations at 20-metre accuracy, close enough to hold a ship inside a dredged channel. They cover the country’s major ports, airports and shipping lanes. A proposal would take the chain to eight stations by 2027 for roughly $200 million.

The international structure is forming around them. A standards working group opened by the United Kingdom, South Korea and France now includes the European Space Agency’s NAVISP engineering team, and it met in Seoul in July.

MagNav: the compass with quantum sensors

Earth’s crust carries a magnetic signature that varies from place to place and stays put. A magnetometer sensitive enough to read it, against a surveyed map, returns a position. The field cannot be jammed or spoofed, because an attacker would have to change the geology.

The Defense Innovation Unit began this in spring 2024 as Transition of Quantum Sensing, drew applications from 72 companies, and selected Honeywell Aerospace to build the prototype that flew the Alaska route. A C-17 Globemaster III demonstration follows.

The software path matters more for commercial fleets. SandboxAQ’s AQNav flew on Northrop Grumman’s Lumberjack in August 2026, installed in under an hour as a software load on existing hardware, which points at retrofit rather than new build.

The sextant as an instrument again

Automated celestial navigation removes the two things that made star sights impractical for routine commercial use: the skill floor and the clear-sky requirement. A star tracker takes sights continuously, day and night, and feeds them to an inertial platform that carries the solution between fixes. Draper’s sliced-lens design reaches 50-metre accuracy in GNSS-denied conditions, against the 3 to 5 metres a civil GPS receiver returns on a clear day: the same order of answer, from starlight. Observing in the near-infrared, or from above most cloud, extends the working envelope.

The pairing is the insight. Inertial platforms drift with time and celestial fixes correct drift, so the combination holds accuracy indefinitely where either alone decays.

At the far end of the same idea, NASA’s SEXTANT experiment demonstrated autonomous X-ray pulsar navigation aboard the International Space Station in 2017, fixing position from millisecond pulsars. The name was chosen well.

The satellite answer that partly works

One category sits outside the old-physics pattern and earns its place on signal strength rather than on independence from orbit.

Low Earth orbit sits roughly twenty times closer to the ground than the GNSS constellations, and signal power rises accordingly. Iridium’s Satellite Time and Location service transmits in L-band at levels about 1,000 times stronger than GNSS, which penetrates buildings and resists jamming that defeats a standard receiver. Xona’s Pulsar constellation targets centimetre accuracy at up to 100 times GPS signal strength, reaching existing receivers through software updates, with its first production spacecraft launched in June 2025. TrustPoint is building a C-band constellation for frequency diversity against the roughly 200 L-band navigation satellites in orbit, with a soft launch targeted for 2027.

The honest caveat belongs in the same paragraph. These are satellites. They raise the power an attacker must produce and the sophistication a spoof requires, and they keep the dependency on orbit that the September 2026 weapons confirmation put in question. LEO PNT is a strong upgrade to the satellite layer rather than an alternative to it, and it belongs in a stack that also holds something on the ground.

Ports have solved position and skipped timing

Container terminals are further along than the rest of logistics, for a reason that predates the threat: GNSS delivers too little accuracy in a yard. Satellite blind zones open up when several quay cranes stand side by side over one ship, so terminals built around other references years ago.

Function	Reference in use
Straddle carrier and crane positioning	LiDAR, laser with fixed reflectors, onboard cameras
Lane, block and stack resolution	Ultra-wideband anchors at centimetre accuracy
Yard equipment tracking	RTK GNSS hybridised with UWB where geometry allows

That is diversity of physics arrived at commercially, without a threat briefing. A terminal running LiDAR and UWB keeps moving boxes through a GNSS denial, because the yard has taken its position from elsewhere all along.

Timing is the gap. Terminal operating systems, gate automation, customs interfaces and equipment telematics synchronise to a clock that mostly comes from GNSS, and the holdover table earlier in this piece is the whole of the defence. A terminal that has engineered position away from satellites and left its timing rack on an OCXO has solved the visible problem and kept the one that stops the software.

Training: the part that equipment cannot buy

Every item above is procurement. Capability requires the people who use it, and this is where the record is weakest.

The evidence is direct. Coast Guard cyber protection teams find default credentials still in place on more than two-thirds of the missions they run, which describes equipment installed and left at factory settings. A January 2025 argument in the U.S. Naval Institute’s Proceedings holds that reinstated orientation lectures fall short of competence, and that celestial navigation requires practice at sea to be real.

Capability	Holds it today	What competence requires
Celestial fix	Deck officers with academy background	Regular sights at sea, worked to a fix
GNSS-denied bridge work	Crews on radar and visual bearings	Drills with the plotter switched off
Spoof recognition	Few, and detection is often absent	Cross-check discipline and alarm literacy
HF radio operation	Concentrated in older officers	Scheduled traffic and exercise use
Manual terminal operations	Gate staff who worked pre-automation	Annual fallback exercises at the berth

Four practices separate an operator that holds capability from one that holds inventory:

1. Run the drill with the system off. A bridge that navigates a coastal passage on radar, bearings and a paper plot once a quarter finds out what it retains, and crews report hours of degradation where others report days.
2. Exercise the manual fallback at the berth. North Carolina Ports moved to manual gate processing in August 2026 and kept cargo moving. That works where procedures exist and staff have used them.
3. Train alarm literacy. Interference detection produces a warning that means switch methods. A crew that treats it as a nuisance alarm has the equipment and none of the protection.
4. Keep the older skills current through use. HF traffic, signal flags and dead reckoning survive where they appear on a schedule rather than in a syllabus.

What it costs, and the order to buy it

Sequencing matters more than the total. The cheap measures cover the common cases, and the expensive ones cover the rare severe case.

Horizon	Measure	Relative cost
This quarter	Multi-constellation receivers; enable interference detection; cross-check configuration; write the GNSS-denied procedure	Low
This quarter	Inventory the oscillator in every timing rack and record its holdover	Low
This year	Upgrade critical-site oscillators from OCXO to rubidium; schedule GPS-denied drills; restore paper chart stock and plotting practice	Moderate
This year	Contract alternative PNT where a service exists, and specify fail-loud behaviour in procurement	Moderate
This decade	Terrestrial reference coverage; magnetic and celestial systems as they commercialise; cesium at national-scale nodes	High, and largely public

The division of labour follows the cost line. Receivers, clocks, procedures and drills sit with operators and pay for themselves against a single avoided outage. Terrestrial chains are public infrastructure, and the reason Britain and South Korea appear in this piece while the United States appears in the mandate is that someone has to build the transmitters.

Specify fail-loud

One procurement clause carries more weight than any other: equipment must announce interference rather than absorb it.

A receiver that accepts a spoofed solution and displays it with full confidence converts an attack into a navigation error the crew will act on. A receiver that flags the inconsistency converts the same attack into a procedure. The hardware cost between those two behaviours is small, and the operational difference is the whole problem.

The same clause applies to timing. A clock that free-runs past its specification without raising an alarm delivers a system that fails quietly, and quiet failure in a timing chain surfaces as data corruption rather than an outage.

What can move the market?

• national funding decisions on terrestrial PNT chains
• the international eLoran standards group and its published specifications
• MagNav progress from military demonstration to commercial certification
• certification pathways for automated celestial systems in commercial fleets
• insurance pricing that distinguishes resilient from unprotected operators
• IMO and ICAO requirements for backup navigation capability
• flag-state and class rules on GNSS-denied competence
• oscillator supply chains for rubidium and cesium references
• alternative PNT service pricing, including satellite-delivered offerings
• receiver chipset adoption of interference detection as standard
• port and terminal investment in independent timing
• exercise regimes that make GPS-denied operation routine

Xin.bz bottom line

The engineering answer to contested orbits exists, and it is mostly old.

Ground-based radio, the magnetic field, the stars and a good clock each return position or time from physics that jamming and spoofing leave untouched. What changed is the sensor and the processing behind them. Quantum magnetometers read the crust, star trackers take sights in daylight, rubidium and cesium references hold a network together for weeks, and receivers recognise a lie.

The cost line is favourable at the operator level. Multi-constellation receivers, interference detection and an oscillator upgrade are capital items measured in thousands, set against a modelled billion dollars a day. The national-scale pieces are the slow ones, and Britain and South Korea are building transmitters while others hold mandates.

Training decides whether any of it works. Equipment installed at factory settings, and crews whose first unplanned outage is their first passage without the plotter, produce a resilient procurement record and a fragile operation.

Redundancy is a question of physics rather than quantity, and the operators who come through the next denial are the ones who bought different physics and practised using it.

Sources

Government and regulatory

• U.K. Ministry of Defence. Urgent Compass eLoran programme, launched May 2026; Chain Odyssey terrestrial resilience network.
• Republic of Korea. eLoran station deployment and coverage of ports, airports and shipping lanes.
• European Space Agency, Navigation Innovation Support Programme (NAVISP). Participation in the international eLoran standards working group, Seoul, July 2026.
• U.S. Defense Innovation Unit. Transition of Quantum Sensing programme, launched spring 2024; MagNav flight demonstration with Honeywell Aerospace, 2026.
• National Institute of Standards and Technology. Economic Benefits of the Global Positioning System (GPS), prepared by RTI International. June 2019.
• U.S. Government Accountability Office. Coast Guard: Additional Efforts Needed to Address Cybersecurity Risks to the Maritime Transportation System, GAO-25-107244. 2025.
• U.S. Coast Guard Cyber Command. Cyber Trends and Insights in the Marine Environment, annual reports.
• International Telecommunication Union. Recommendations ITU-T G.8272 and G.8272.1, primary and enhanced primary reference time clocks.

Academic and research

• NASA Goddard Space Flight Center. Station Explorer for X-ray Timing and Navigation Technology (SEXTANT): autonomous X-ray pulsar navigation demonstration, 2017.
• Draper Laboratory. Sliced-lens star tracker celestial navigation patents and published accuracy results.
• U.S. Naval Institute. Ships Must Practice Celestial Navigation, Proceedings 151/1/1,463. January 2025.

Industry

• Honeywell Aerospace. MagNav prototype flight test results, 2026.
• SandboxAQ. AQNav flight test aboard Northrop Grumman Lumberjack, August 2026.
• Oscillator and timing vendors. Published holdover specifications for OCXO, rubidium and cesium references.