Argues that NATO, EASA, and national regulators have spent two years treating the Baltic GNSS interference as either operational noise or politically too sensitive to name in public technical documents. The paper's value isn't novel science — it's the boring work of stitching together TDOA measurements, NIC=0 aircraft events, and Galileo SNR anomalies into a public, CC-BY-licensed receipt that risk teams can actually cite.
Submitted the arXiv paper to Hacker News where it drew 377 points and 200 comments, signaling strong community interest in surfacing this kind of public-domain technical analysis. The submission framing — 'Tracing a powerful GNSS interference source over Europe' — treats the triangulation itself as the newsworthy act.
Highlights that the authors cross-reference multi-station TDOA against aircraft NIC=0 reports, ground monitoring receiver logs, and Galileo SNR anomalies harvested from open networks, then reproduce the effect on the ground with consumer-grade SDRs. The unglamorous, auditable nature of the dataset — tens of thousands of aircraft hours, tight great-circle bearing clustering — is what makes the conclusion hard to dismiss without producing counter-evidence.
Notes the authors give coordinates and power estimates but deliberately avoid flag-planting attribution language, even though the location is 'one of the worst-kept secrets in European aviation safety circles.' This restraint keeps the paper in the realm of verifiable signal geometry rather than geopolitical claim-making, which is exactly what makes it usable by airline risk teams and telco timing engineers who need defensible technical references.
A paper posted to arXiv this week (2606.03673) does something that NATO, EASA, and the various national civil aviation regulators have conspicuously avoided doing in public for two years: it triangulates the geographic source of one of the strongest GNSS interference emitters affecting European airspace. The methodology is unglamorous in the best way — multi-station time-difference-of-arrival (TDOA) measurements cross-referenced against aircraft-reported Navigation Integrity Category (NIC=0) events, ground monitoring receiver logs, and Galileo signal-to-noise ratio anomalies harvested from open networks.
The conclusion is the kind of thing intelligence services normally publish in a classified annex, not on arXiv with a CC-BY license: a single, persistent, high-power emitter near the eastern Baltic accounts for the dominant share of L1-band interference detected across the region. The paper carefully avoids attribution language — it gives coordinates and power estimates, not a flag — but the location is one of the worst-kept secrets in European aviation safety circles.
The scale matters. The dataset spans tens of thousands of aircraft hours over Finland, Estonia, Latvia, Lithuania, and Poland, with NIC=0 events (essentially, the aircraft telling ATC "my GPS is unusable") clustering tightly around great-circle bearings from the emitter. The authors also reproduce the effect on the ground using consumer-grade software-defined radios. None of this is novel science. What's novel is that someone finally did the boring work of stitching the receipts together and posting them publicly.
For most of the last decade, GNSS interference reports in the Baltic were treated as either operational noise — "the pilots will deal with it" — or as a politically sensitive topic too hot to name in a public technical document. That ambiguity has been load-bearing for a lot of bad threat models. Risk teams at airlines, telcos, and timing-sensitive financial venues have been able to wave the issue away as "localized" or "intermittent." The new paper makes both descriptions untenable.
Once the source is a point on a map with a measured ERP, the problem stops being meteorological and starts being engineering — which means it has a budget line. That's the real shift here. You can't procure mitigation for vibes; you can procure it against a documented adversary emitter with a known frequency profile and a known duty cycle.
The second-order effect is reputational. EUROCONTROL has been publishing aggregate interference dashboards for over a year, but always with the locations redacted to grid squares. Civilian researchers publishing un-redacted localizations on arXiv put pressure on the official channels to either confirm, deny, or explain the asymmetry. Expect the next round of EASA Safety Information Bulletins to be noticeably less coy.
There's also a community-reaction angle worth noting. The HN thread, currently at 377 points, is overwhelmingly populated by people who actually own this problem operationally — Part 91 pilots, maritime engineers, telecom RF leads, and the small but vocal cohort of timing-receiver vendors. The technical comments are uniformly "yes, we've been seeing this since 2023; thanks for finally writing it down." The political comments are doing what political comments always do. The signal is in the first group.
The most useful technical observation in the thread is that the interference is consistent with a relatively simple high-power CW or swept-CW jammer, not a sophisticated spoofer — which is good news for detection and bad news for anyone hoping the problem will solve itself. Jammers are loud and dumb; spoofers are quiet and surgical. The current emitter is the former, but the existence of the paper materially raises the incentive for the operator to upgrade to the latter.
If your only exposure to GPS is the location chip in your phone, this is interesting reading and nothing more. If you operate any of the following, your weekend reading list just got longer:
Timing-sensitive infrastructure. NTP stratum-1 servers backed by GPS disciplined oscillators are the single most common production dependency on GNSS that engineers forget they have. If you run telco 5G synchronization, financial trading colocation, broadcast playout, or any PTP grandmaster, you should already have a holdover oscillator spec and a documented behavior for what happens when the GPS lock drops. If your runbook for "GPS lock lost" is "page someone and hope," the new paper is your excuse to fund a chronometer-grade OCXO or a Rb holdover, plus a secondary time source (PTP from a fiber peer, eLORAN where available, or a second receiver on Galileo with OSNMA enabled).
Fleet and logistics. Dead reckoning from IMU plus map-matching gets you surprisingly far when GPS drops, but only if the integration was designed for it. Most off-the-shelf telematics stacks treat GPS loss as a transient and silently extrapolate. The fix is multi-constellation receivers (GPS + Galileo + BeiDou + GLONASS) with per-constellation health monitoring, not just "more satellites."
Drones and autonomy. If you're shipping anything autonomous that flies, drives, or sails near the Baltic, multi-constellation alone is not enough — you need anti-spoof receivers (Galileo OSNMA is the cheapest path; military M-code is the most expensive) and a sensor-fusion stack that can degrade gracefully to visual-inertial odometry. The fact that the current emitter is a jammer rather than a spoofer is a grace period, not a guarantee.
Mobile and web devs. You're probably fine, but be aware that user location can degrade dramatically in affected regions for reasons that look like a bug in your code. If you have users in Finland or the Baltic states reporting wildly wrong locations, it's not your geocoder.
The arms race here is depressingly predictable. Public triangulation raises the cost of operating a jammer in the open, which historically drives operators toward either mobile emitters (harder to localize) or spoofers (harder to detect at all). Galileo OSNMA, which authenticates the navigation message itself, is rolling out across consumer chipsets through 2026 and 2027 — that's the structural mitigation, and it's coming faster than most people realize. In the meantime, the playbook is unchanged from what aviation has been quietly doing for two years: assume GNSS is best-effort, design holdover, and stop treating PNT as a free utility. The paper just gave that argument receipts.
Related Veritasium video: https://www.youtube.com/watch?v=tz23G_UXCGA
Mildly interesting, and highly likely related. A cluster of 5(?) Ukrainian marine drones wound up today outside and around of Constanta off the coast of Romania with one detonating in the port and the rest detonating... somewhere around. Que here noisy exposion in port:https://youtu.be
TLDR (conclusion from the paper): "By a combination of these techniques the satellite Cosmos 2546 (NORAD ID 45608) was identified with high confidence as one source of the interference. Further analysis pointed to the Russian Edinaya Kosmicheskaya Sistema, an early warning constellation to whic
I wonder why they call this specific discovery “jamming”. What they found is a relatively rare burst transmissions over roughly 5MHz of spectrum of something looking like a 12ms cyclic prefix with spacing related to 150 seconds multiplies. I would suspect it is some sort of sync or data close to L1
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
Interesting to see that they are able to identify the specific satellite. I wonder if we can do something now that we know the source.Working on construction projects on the Romanian coastline (just South of Ukraine) and on the Polish continental waters (just West of Kaliningrad) we experienced jamm