Key takeaways
- ADS-B position reports are broadcast in the clear, with no authentication, and they inherit whatever GNSS tells the aircraft. Both jamming (signal loss) and spoofing (false position) show up in the surveillance picture that controllers, collision-avoidance logic and other aircraft rely on.
- In 2023 data covering 376.9 billion messages, I found 53,232 GPS gaps and 5.9 million GPS deviations. Gaps in areas covered by an active navigation NOTAM lasted on average 257 seconds longer, and many affected messages still carried a "good" integrity flag (NIC above 7).
- Anomalies are not random noise. In 2025 data, 98.9% of high-utilization aircraft showed temporally clustered anomalies, with sustained bursts averaging 28.83 seconds, far beyond the 5-second fault tolerance of typical decentralized trackers.
- Interference density in a region is a statistically significant predictor of conflict events in that region. The threat map is not static; it follows geopolitics.
- Encrypting ADS-B is not a realistic near-term fix. The practical defences are plausibility checks, independent position sources, procedures, and knowing your own exposure by route and season.
Why this matters more every year
Automatic Dependent Surveillance–Broadcast (ADS-B) is the primary cooperative surveillance link of modern air traffic management. Each equipped aircraft broadcasts its GNSS-derived position, velocity and identity roughly once a second on 1090 MHz, and everyone in range, including ground stations, other aircraft, hobbyist receivers and crowdsourced networks, can listen. Programs like NextGen and SESAR, and safety systems like the Airborne Collision Avoidance System X (ACAS X), are built on the assumption that these reports are timely and true.
Two design decisions made in the 1990s and 2000s now define the risk. First, ADS-B messages carry no cryptographic authentication: any transmitter can claim any position. Second, the position itself comes from GNSS, which is a weak signal from space that can be drowned out (jamming) or replaced by a stronger fake (spoofing) with commodity equipment. ICAO, IATA, EASA and the FAA have all named GNSS interference a critical threat to civil aviation, and since 2022 the volume of reported interference in the Baltic, Black Sea, eastern Mediterranean and Middle East has grown from a curiosity into a daily operational problem.
For a small operator, this shows up as unexpected map shifts, spurious terrain or traffic warnings and RAIM alerts. For a UAS or eVTOL developer, it is a certification and design question: what does the aircraft do when its primary position source lies?
Gaps, deviations and ghost records: what the data actually shows
My work uses three observable signatures in ADS-B data. They matter because they map onto different failure modes and different defences.
| Signature | What it looks like in the data | Most likely causes |
|---|---|---|
| GPS gap | An aircraft keeps transmitting but its position fields go missing or the track drops out for a period, then resumes. | Jamming, receiver loss of lock, coverage holes, equipment faults. |
| GPS deviation | Reported position jumps in a way that is physically impossible: "teleportation" between successive reports, altitude sources that disagree, velocity that does not match the movement of the track. | Spoofing, multipath, receiver errors, decoding artefacts. |
| Ghost record | A message that reaches the network with positional coordinates but with other telemetry partially missing. Traditional pipelines discard these before anyone looks at them. | Severe transponder or receiver degradation; a signature that standard filters were hiding. |
Three studies, each on a full year of global data from the OpenSky Network, give the numbers below. The full papers are on the Research page.
Evidence · 2023 baseline
GPS Anomalies in Aviation, Journal of Aerospace Information Systems, 2025. From 376.9 billion ADS-B messages we identified 53,232 GPS gaps in 37,546 distinct flights and 5,878,275 GPS deviations from 101,681 aircraft, and compared them with 455,385 US and 30,160 international NOTAMs. Findings: anomalies concentrate where traffic is dense; gaps inside active navigation-NOTAM windows last on average 257 seconds longer; about 19% of the top anomaly cells overlapped NOTAM cells, all of them in North America; solar flux did not correlate with monthly gap counts; and a significant share of messages with missing coordinates still reported a Navigation Integrity Category above 7.
Evidence · 2025 baselines
Global Traffic-Normalized Baseline for ADS-B Kinematic Integrity and Ghost Records (with J. S. D. Garcia and T. A. Smith, 2026) normalizes anomaly counts by flight hours across 260 Flight Information Regions using 246.7 billion messages and 77.5 million flight hours. It identifies 3,818,743 ghost records, shows that several densely instrumented European FIRs recorded zero severe breaches in the whole year, and independently flags airspaces such as Tripoli, Lahore, Baku and Amman as kinematically volatile, matching reported interference patterns.
Degradation of ADS-B Integrity: Temporal Clustering and Anomaly Persistence (Reliability Engineering & System Safety, 2026) follows 229.0 billion state vectors at the level of the individual aircraft. Anomalies cluster in time for 98.9% of aircraft flying 20 or more hours a month, with a characteristic 3-second scale and a mean sustained duration of 28.83 seconds. 58.4% of the fleet recorded clustered anomalies on two or more days. Recurrence is linked more strongly to how well the aircraft is observed by the receiver network (hazard ratio 1.52) than to how much it flies (1.19).
What those numbers mean in practice
- Integrity flags are not enough. NIC and similar quality indicators describe what the receiver believes about its own solution. A spoofed receiver believes the spoof. Your monitoring has to check physics (can this aircraft really have moved that far in one second?), not just flags.
- Bursts outlast the tolerance of downstream systems. A tracker that tolerates five seconds of bad data will be fed almost thirty seconds of it in a typical burst. Anything that extrapolates position for conflict detection or metering is exposed.
- Where you are observed matters. Recurrence tracks network observability. Regions with dense, redundant ground reception both suffer fewer severe breaches and catch more of the minor ones. Sparse regions hide problems.
- The map moves. In Predicting geopolitical instability through GNSS anomalies and air traffic data (Journal of Transportation Security, 2025), GNSS deviation density was a positive, statistically significant predictor of conflict-event counts in every model specification we tested, across 23,665 observations. Interference is a leading edge of geopolitical risk, so last year's exposure map is not this year's.
Why "just encrypt ADS-B" isn't the answer
An older version of this site listed encryption of ADS-B signals as an essential measure. I no longer say that, and here is why.
- The installed base. Tens of thousands of transponders certified to DO-260B are flying today, and the 1090ES message format is fixed by international standard. Changing the format is a multi-decade, multi-authority program, not a product feature.
- Encryption solves the wrong problem. The threat to safety is not eavesdropping; ADS-B is meant to be received by everyone. The threat is false reports and absent reports. Authentication (proving a message came from the aircraft it claims) is the relevant property, and even that does nothing against jamming or against a genuine aircraft honestly rebroadcasting a spoofed GNSS position.
- Key management at global scale. Any authentication scheme needs keys distributed and revoked across every state, operator and receiver. Research proposals exist (format-preserving schemes, out-of-band signatures, TESLA-style delayed keys), and they are worth following, but none is on a certification path for the current fleet.
The realistic defences are the ones that assume the signal can be wrong and detect it: independent position sources (secondary radar, wide-area multilateration, inertial reference), physical plausibility filters on the reports themselves, cross-checks between altitude sources, and procedures that keep crews and controllers effective when the picture degrades.
What operators should do
Most of this is procedural and cheap. The expensive part is discovering it during an event.
1. Know your exposure by route and season
Interference is regional and it changes. Map the FIRs and corridors you fly against a current anomaly baseline, not against last year's headlines. Where the data shows persistent gap or deviation density, plan for degraded navigation as the normal case, not the exception.
2. Brief the failure modes, not just the warning
Crews should know what spoofing looks like from the flight deck: position shifts that do not match ground speed, sudden time or date changes on GNSS-fed clocks, unexpected terrain or traffic alerts, and RAIM or "GPS unavailable" messages that appear and clear in bursts. EASA's Safety Information Bulletin on GNSS outages and alterations (SIB 2022-02, latest revision) is a good basis for a briefing sheet.
3. Rehearse the revert
Conventional navigation, inertial drift limits, and coordination with ATC when GNSS is unreliable are perishable skills. A short recurrent exercise costs less than one diversion.
4. Treat navigation NOTAMs as the signal they are
The 2023 data showed gaps lasting longer where an active navigation NOTAM was in force. NOTAMs are imperfect, but they carried real information about where interference was happening. Make sure they reach the crew and the dispatcher in a form they will read.
5. Report
Report suspected spoofing and jamming through your national occurrence reporting channel and to the ANSP. Aggregated reports are how NOTAMs and advisories get issued for the next crew.
6. Extend the thinking to the ground
Timing for many ground systems (voice recorders, logging, some maintenance and dispatch tools) comes from GNSS. A spoofed time source can corrupt records and confuse investigations. Know which of your systems depend on it.
What UAS and eVTOL developers should do
For a new aircraft, GNSS interference is a threat condition to be designed against, not a nuisance to be tolerated. The airworthiness security process in DO-326A/ED-202A expects you to identify it, assess its severity and show that your measures work. In practice, that means:
- Never let GNSS be a single point of truth. Fuse it with inertial sensing and, where the mission allows, with barometric, visual or radio-based positioning. Define what the flight controller does when sources disagree beyond a physical bound.
- Apply the same plausibility logic my research uses to the aircraft itself. A position jump that implies hundreds of metres per second, a geometric altitude that diverges from barometric, a velocity that contradicts the trajectory: these are computable in flight, in real time, from the data the aircraft already has.
- Make loss of GNSS a designed, tested behaviour. Hover-and-hold, return-to-home and geofencing all fail in interesting ways when position is wrong rather than absent. Test the "wrong" case, not just the "absent" one.
- Keep the command-and-control link independent of GNSS. If your C2 link timing, frequency hopping or authentication depends on GNSS time, a spoofed clock can take down control along with navigation.
- Remember that Remote ID is also unauthenticated. Broadcast Remote ID inherits the same design assumption as ADS-B: anyone can transmit anything. Do not build safety logic that trusts it blindly.
- Design for the ground segment too. The ground control station, its GNSS receiver and its network connection are part of the aircraft system for certification purposes.
A short checklist
| Question | If the answer is no |
|---|---|
| Do you have a current anomaly baseline for the regions you operate in? | Get one before the next season; the picture moves with geopolitics. |
| Can your crews describe what spoofing looks like from the flight deck? | One page and a ten-minute briefing fix this. |
| Have you flown the "GNSS is lying" case in the simulator or the test range? | Add it to recurrent training or the flight-test plan. |
| Do you know which ground systems take time from GNSS? | Inventory them; add an independent time source where records matter. |
| For developers: is GNSS interference in your threat model with a severity and a measure? | It will be one of the first questions in a DO-326A review. |
Questions I get asked
Can ADS-B tell me whether my aircraft is being spoofed or jammed?
Not by itself. A jammed receiver usually shows up as a gap in position reports; a spoofed one as a position that jumps in a way the aircraft could not fly. Both are visible after the fact in the surveillance picture, and the integrity flag in the message often still reads as good. The reliable signs are disagreement between GNSS and an independent source such as inertial, DME/DME or radar, and kinematics that are physically implausible.
Where is GNSS interference most likely on my routes?
Interference follows geopolitics. In 2023 data the regions with the most gaps and deviations sat near active conflicts and military exercises, and interference density was a statistically significant predictor of conflict events. Check the navigation NOTAMs for your route and season, and expect the map to change from year to year.
Wouldn't encrypting ADS-B solve this?
No. Encryption would require every aircraft and ground station in the world to change at once, it would close the open air-traffic picture that collision avoidance and safety research depend on, and it would do nothing about GNSS, which is the upstream source of most anomalies. Plausibility checks, independent position sources and procedures work today.
How much of the ADS-B data is actually affected?
In 376.9 billion messages from 2023 there were 53,232 gaps and 5.9 million deviations: a small fraction of messages, but concentrated in particular regions and periods. In 2025 data, anomalies clustered in time for 98.9% of high-utilization aircraft, in bursts averaging 28.83 seconds, well beyond the 5-second tolerance of typical trackers.
How I can help
I offer a GNSS and ADS-B integrity analysis for the regions and routes you fly, built with the same global baselines as the research above: maps and rates of gaps, deviations and surveillance anomalies for your operating area, with recommendations for procedures or, for developers, for the aircraft's design and threat model.
Ask about an exposure analysis
Founder of Mevocopter Aerospace and Ph.D. researcher in Data Science at the University of Essex, working on kinematic integrity and security in automated aviation systems. Lead author of the GET framework and the open GET-25 ADS-B benchmark. Publications · ORCID
References
- Pik, E., Berra, M., Yearwood, J., Garcia, J. S. D. (2025). GPS Anomalies in Aviation: Preliminary Insights from Automatic Dependent Surveillance–Broadcast Data. Journal of Aerospace Information Systems, 22(7), 570–582. DOI: 10.2514/1.I011527 · PDF (accepted manuscript)
- Pik, E., Garcia, J. S. D., Smith, T. A., Kocaman, I., Berra, M. (2025). Predicting geopolitical instability through GNSS anomalies and air traffic data. Journal of Transportation Security, 18, 29. DOI: 10.1007/s12198-025-00323-w · PDF (accepted manuscript)
- Pik, E. (2026). Degradation of Automatic Dependent Surveillance-Broadcast integrity: Temporal clustering and anomaly persistence. Reliability Engineering & System Safety, 113534. DOI: 10.1016/j.ress.2026.113534 · Accepted manuscript (PDF)
- Pik, E., Garcia, J. S. D., Smith, T. A. (2026). Global Traffic-Normalized Baseline for ADS-B Kinematic Integrity and Ghost Records. Manuscript. Data and code: GET-25 benchmark
- Pik, E., Berra, M., Yearwood, J., Garcia, J. S. D. (2024). Detecting GPS Anomalies in Aviation Using ADS-B: Correlating Coordinate Gaps and GPS Deviations with NOTAM Warnings. AIAA AVIATION Forum 2024. DOI: 10.2514/6.2024-4640 · PDF
- RTCA DO-326A / EUROCAE ED-202A, Airworthiness Security Process Specification; RTCA DO-260B, Minimum Operational Performance Standards for 1090 MHz Extended Squitter ADS-B.
- EASA Safety Information Bulletin 2022-02 (latest revision), Global Navigation Satellite System Outage and Alterations Leading to Communication/Navigation/Surveillance Degradation.