ELINT · Electronic Intelligence

Twelve hours without GNSS: the UK’s resilient-timing market test makes holdover the requirement

Dstl has asked industry for a deployable system that keeps time within 30 nanoseconds of UTC through twelve hours with no external signal at all. The specification says more about the electromagnetic environment planners now expect than any interference report.

The requirement is the finding

The UK Defence Science and Technology Laboratory (Dstl) has asked industry for a “resilient timing and positioning system for operation in degraded GNSS environments” [1]. The request for information was reported on 12 November 2026. Responses are due by 17:00 on 9 December [1].

It is market research. Dstl reserves the right not to procure anything [1]. The value for practitioners is the specification. It states, in numbers, what a defence laboratory now expects a deployable timing source to survive.

What Dstl asked for

The system must supply two outputs to an external system operating in contested electromagnetic environments: a UTC-referenced timing signal and a static position [1]. The core figures are these [1].

  • Timing better than ±30 ns relative to UTC, with a 1 PPS output and a 10 MHz reference.
  • Static position better than 5 m against surveyed ground truth.
  • Both held through a continuous 12-hour loss of all external PNT sources.
  • Performance maintained under GNSS jamming and spoofing, including first acquisition under those conditions.
  • Technology at TRL 9, or evidence of equivalent operational deployment.

The exclusions matter as much. The solution must not depend on fixed local infrastructure such as fibre timing distribution, on user-generated PNT signals, on ad hoc reference sources, or on ITAR-controlled components [1]. Any non-GNSS source, LEO-PNT included, must itself be resilient to jamming and interference [1]. Respondents must describe their oscillator technology, naming OCXO, DOCXO, rubidium and chip-scale atomic clocks as examples, and its contribution to holdover [1].

How tight thirty nanoseconds is

The telecom sector has the closest published benchmark. ITU-T Recommendation G.8272 defines two classes of primary reference time clock. Class A must stay within 100 ns of UTC. Class B must stay within 40 ns [6][7]. Both figures assume normal, locked operation [6].

The enhanced primary reference time clock, G.8272.1, is the more demanding class. One industry summary of the standard puts its locked accuracy at 30 ns, and its holdover at 100 ns for up to 40 days [8]. The ITU’s own summary does not quote a number. It says only that the enhanced clock can sustain primary-reference performance for weeks without GNSS [6]. The two descriptions differ in precision, not direction.

Read against that scale, Dstl wants the enhanced class’s locked accuracy held while unlocked for twelve hours. It wants it in a system that cannot lean on a national fibre network. That combination puts the burden on the local clock and on whatever non-GNSS signal the supplier can prove is hard to jam.

Why the exclusions point in one direction

Civil timing resilience has leaned on terrestrial distribution. The UK’s own 2023 policy framework proposes a sovereign National Timing Centre and a defence timing service built on it [10]. South Korea is moving the same way. Its standards institute, KRISS, plans to carry national time over the KREONET research network, has held national time within 5 ns of UTC over the past year, and has shown picosecond-level transfer over a 230 km link [3].

Dstl rules that approach out for this use case. The plain reading is that the equipment must work where no such network exists. A field system carries its time with it.

The LEO-PNT clause is the second signal. Commercial low-orbit navigation is close. Xona Space Systems launched six Pulsar spacecraft in late October and plans early-user service in mid-2027 [5]. Its first customers will get eight broadcasts a day of about 25 minutes each, and the initial satellites give a revisit time of about 12 hours [5]. Xona’s claims that Pulsar outperforms GNSS under jamming are its own and were not independently tested in the reporting [5].

A twelve-hour revisit and a twelve-hour holdover requirement fit together. In early service, a LEO-PNT signal will discipline a clock between passes. It will not replace the clock. Dstl’s demand that such sources also resist jamming means a supplier cannot treat them as a clean fallback.

The environment behind the numbers

Finland’s regulator, Traficom, publishes the clearest open series. It recorded 61 aviation GPS interference reports in 2022, 236 in 2023 and 2,800 in 2024 [2]. Reporting criteria changed on 1 May 2024, so 2024 is not directly comparable with earlier years [2]. The count fell to 1,704 in 2025 and reached 2,451 by the end of October 2026 [2]. Traficom says interference in southern Finland has continued daily since January 2024 [2].

Its maritime figure rose to 406 in the first ten months of 2026, against 231 in 2025 [2]. That comparison is weak. From January 2026 Traficom counts affected vessels rather than reports [2].

Interference is not only adversarial. In the United States, a medical evacuation aircraft struck a ridge near Ruidoso, New Mexico, on 14 May, killing four people [11]. The NTSB’s preliminary report describes an abrupt loss of GPS linked to active military electronic warfare testing; no probable cause has been determined [11]. The GPS Innovation Alliance has since called for tighter coordination of government test activity [11].

What a holdover requirement buys

The UK government’s own economics show where holdover pays. A study published in 2023 estimated that a UK-wide 24-hour GNSS outage would cost about £1.4 billion [9][10]. For a seven-day outage the government release gives £7.6 billion [10]. The study itself gives £7,644 million in one place and £7,625.3 million in another [9]. The same study expects no loss in finance over seven days, because trading systems carry holdover oscillators [9]. It estimates resilient PNT could cut the seven-day loss by almost half [9].

Authentication solves a different problem. Japan’s QZSS plans to start its QZNMA signal-authentication service in late February 2027, including authentication data for GPS and Galileo signals [4]. Authentication helps a receiver reject a forged signal. It does not supply time when the band is jammed. Holdover does.

What to do

  • Write your holdover figure down. For every timing-dependent service, state the accuracy it needs and how long the local oscillator can hold it with GNSS absent [1][9].
  • Test it. Disconnect the antenna on a reference unit and measure drift against an independent source, rather than relying on datasheet values [1].
  • Map your timing to the ITU-T classes. Know whether a site needs 100 ns, 40 ns or better, and whether its clock can hold that through a realistic outage [6][7].
  • Treat new signals as inputs to the clock, not substitutes for it. Ask LEO-PNT suppliers for revisit times and jamming evidence, not only accuracy [1][5].
  • Add authentication where it is offered, as a spoofing control, without counting it as continuity [4].
  • Track regulator data, not anecdotes. Traficom’s series shows how counting rules change the trend [2].

The Dstl request is not a purchase. It is a public statement of a planning assumption: hours without satellite time, under jamming, with no wire to fall back on. Operators across APAC who run telecoms, grid or financial timing on GNSS should test their own systems against the same assumption before an outage does it for them.

Sources

Every R3KONX article cites its primary material. 11 sources, in order of first citation. Links open the original publication.

  1. U.K. tender on resilient PNT GPS World · 2026-11-12
  2. Satellite navigation service interferences in Finland Traficom (Finnish Transport and Communications Agency) · 2026-11-12
  3. Korea to build precision time and frequency network to reduce reliance on GNSS GPS World · 2026-11-10
  4. Notice on the start of services from QZS-7 (Michibiki 7) Cabinet Office, Government of Japan (QZSS) · 2026-11-13
  5. 6 Pulsar satellites launched into LEO, marking the start of Xona's production constellation Inside GNSS · 2026-10-28
  6. PRTC and ePRTC (ITU-T G.8272 and G.8272.1) flyer ITU-T Study Group 15 · 2023
  7. ITU-T G.8272 (PRTC) one-page summary Calnex Solutions · accessed 2026-11-18
  8. ITU-T G.8272.1 (ePRTC) one-page summary Calnex Solutions · accessed 2026-11-18
  9. The economic impact on the UK of a disruption to GNSS: executive summary UK Space Agency / London Economics (GOV.UK) · 2023-10-18
  10. Critical services to be better protected from satellite data disruptions through new position, navigation and timing framework Department for Science, Innovation and Technology (GOV.UK) · 2023-10-18
  11. GPS jamming risks to civilians from military testing DefenseScoop · 2026-11-06

Researched and written by the R3KONX Analysis Desk from the cited primary material. The official Find a Tender notice could not be retrieved; the Dstl requirement is taken from GPS World's reproduction of it. The request is market research, not a contract, and no supplier is assessed here. Corrections: event@r3konx.asia.

R3KONX 2027

Work in this domain? So does the programme

4–6 May 2027, World Trade Centre Kuala Lumpur. Curated talks, hands-on training and The BattleGrid.