ADS-B security: how aircraft signals are attacked, and detected
ADS-B is the backbone of modern air-traffic surveillance, yet it was designed for safety and openness, not security: its messages are not authenticated and not encrypted. This is the complete, plain-English reference to the resulting attack surface, every major attack class, the surge in real-world GNSS spoofing, and the concrete ways these attacks are detected. Maintained alongside an open, citable benchmark dataset.
Attacks on air-traffic signals are no longer theoretical
For a decade, ADS-B attacks lived mostly in security papers. That changed. Industry monitors now report a steep, sustained rise in GNSS (GPS) spoofing affecting civil aviation. Aireon observed roughly a 500% increase in spoofing activity, with on the order of 1,500 flights per day affected in 2025 versus a few hundred in early 2024, and EASA recorded about 6,000 spoofing events in 2024, a quarter of them during approach. Activity clusters near conflict zones but is no longer confined to them.
Figures as reported by Aireon and EASA; see references below. AeroScope is an independent observer-relative platform, it does not control air-traffic systems; it makes these signals legible and flags the inconsistencies attacks leave behind.
Why ADS-B is insecure by design
ADS-B was standardised to make aviation safer and more open, every aircraft continuously announces where it is so everyone can see it. Security was not a design goal. Three properties create the entire attack surface:
- No authentication. A receiver cannot verify who sent a message. Anyone with a transmitter can emit a well-formed ADS-B frame.
- No encryption. Messages are broadcast in the clear, so anyone can receive, record and replay them.
- No integrity protection. Nothing cryptographically binds the contents to a real airframe; the quality fields (NIC/NACp/SIL) describe claimed accuracy, not authenticity.
Researchers demonstrated practical attacks over a decade ago using inexpensive software-defined radios. Schäfer, Lenders & Martinovic showed low-cost injection of fake messages in 2013, and Strohmeier, Lenders & Martinovic mapped the protocol's security in their 2015 IEEE survey. The hardware has only gotten cheaper since. How ADS-B works →
Every major class of ADS-B attack
| Attack | What it does | Effect on the picture |
|---|---|---|
| Eavesdropping | Passively receive all traffic in range | Privacy / reconnaissance; no signal change |
| Jamming | Flood 1090 MHz with noise | Denial of service, aircraft vanish from a whole area |
| Message injection (ghost aircraft) | Transmit fabricated frames for non-existent aircraft | Phantom targets appear; controllers/operators chase nothing |
| Message deletion | Cancel a target's frames via destructive interference | A real aircraft disappears from surveillance |
| Message modification | Alter fields of a real aircraft's broadcasts | Virtual trajectory change, wrong altitude, position or identity |
| Replay | Re-broadcast previously recorded messages | Plausible but stale/duplicated tracks |
| Flooding / DoS | Saturate receivers with high-rate valid-looking frames | Processing overload; real targets crowded out |
| GNSS / GPS spoofing | Fake the satellite signals the aircraft navigates by | The aircraft honestly reports a false position via ADS-B |
The first seven attack the ADS-B link directly. The eighth. GNSS spoofing, is different and increasingly the dominant real-world threat, so it deserves its own section.
ADS-B spoofing vs GNSS spoofing
ADS-B spoofing
The attacker forges the radio message. The aircraft (real or fabricated) "says" something false because someone else transmitted it. Defended by checking message integrity and cross-validating against independent receivers.
GNSS / GPS spoofing
The attacker forges the satellite signal. A real aircraft's GPS computes the wrong position, and the aircraft then broadcasts that wrong position truthfully over ADS-B. This is the surge driving 2024–2025 incidents, often near conflict zones.
Drones, Remote ID and the low-and-slow gap
Drones widen the attack surface from two directions. A compliant drone broadcasting Remote ID can have that broadcast spoofed just like ADS-B. And most small drones broadcast nothing on these frequencies at all, a deliberate-intrusion blind spot that ADS-B alone cannot close. AeroScope flags Remote-ID and low-and-slow candidates honestly, and is explicit that a fully silent drone is invisible to cooperative surveillance. Drone detection →
How these attacks are actually detected
Because ADS-B cannot be authenticated cryptographically (today), defense is about cross-checking every broadcast against things an attacker cannot easily fake at once: physics, message-integrity metadata, and independent geometry.
Integrity-field checks
A DO-260B-style validation of NIC/NACp/NACv/SIL, a fabricated message often gets these wrong or claims precision its track can't support. Integrity model →
Physical plausibility
A Kalman filter predicts the next position; spoofs that teleport, freeze or exceed the performance envelope fail the normalized-innovation test.
Self-consistency residuals
Geometric-vs-barometric altitude, ground-speed-vs-Mach and track-vs-heading should agree on a real airframe; contradictions betray fabrication.
Multi-receiver cross-check
A genuine aircraft is heard by many independent receivers; multilateration and inter-network agreement expose single-source or geometry-inconsistent signals. Multi-source fusion →
Consensus anomaly detection
Six independent torch-free detectors (River, IsolationForest, pykalman, OpenAP, stumpy, PyOD) vote; a flag fires only on agreement. Detectors →
Transparent attention ranking
The signals roll into an explainable 0–100 attention score so a human looks at the right aircraft first, a ranker, never an automated accusation. Threat scoring →
This is precisely what AeroScope runs on every aircraft, every cycle, fused across public receiver networks. See the end-to-end method on how to detect ADS-B spoofing.
What you can actually do
- Watch your own sky. Set your location on the live map and let the integrity and attention layers surface anything inconsistent overhead.
- Treat single-source tracks with caution. An aircraft seen by only one receiver, with jumpy kinematics, deserves a second look, see is that aircraft a threat?
- Know the limits. No public, passive platform can see jamming victims that have gone dark, transponder-off aircraft, or silent drones. Honesty about coverage is part of the defense.
- Report real safety concerns to the authorities. AeroScope helps you notice and document; it is situational awareness, not an interdiction system.
An open benchmark for ADS-B attack detection
Detection research has long been held back by the lack of shared, labelled, openly-licensed attack data. AeroScope publishes the ADS-B Anomaly Benchmark v1 (CC-BY 4.0): real airborne traffic paired with synthetically injected attacks from the standard taxonomy (altitude/velocity tampering, ghost kinematics, integrity-field downgrade; methodology after Habler & Shabtai, 2018), across 38 documented columns, with a reproducible IsolationForest baseline (ROC-AUC 0.87, recomputed from the 459 real and 459 injected rows in the published file). Use it to build and compare your own detectors, also on Hugging Face. Dataset & card →
Sources
· Strohmeier, M., Lenders, V. & Martinovic, I. (2015). On the Security of the Automatic Dependent Surveillance-Broadcast Protocol. IEEE Communications Surveys & Tutorials.
· Schäfer, M., Lenders, V. & Martinovic, I. (2013). Experimental Analysis of Attacks on Next Generation Air Traffic Communication. ACNS.
· Habler, E. & Shabtai, A. (2018). Using LSTM encoder-decoder algorithm for detecting anomalous ADS-B messages. Computers & Security. arXiv:1711.10192
· EASA (2024) reporting on GNSS spoofing/jamming events affecting civil aviation.
· Aireon (2025). Observations of trends in GPS anomalies affecting aviation (white paper).
· OPSGROUP (2025). GPS Spoofing Workgroup. Final Report.