Airspace Awareness from Open Data

§01Executive summary

Most commercial aircraft now broadcast their own position, velocity, identity and navigation quality several times per second, in the clear, on frequencies anyone can receive. In the United States, aircraft operating in the airspace defined in 14 CFR 91.225 have been required to carry ADS-B Out since January 1, 2020, and must transmit at all times unless the FAA or ATC authorizes otherwise²Source 2 · Cornell Law School Legal Information Institute (US Code of Federal Regulations)14 CFR § 91.225 — ADS-B Out equipment and uselaw.cornell.edu. Volunteer and research receiver networks collect these broadcasts at scale; the OpenSky Network was founded precisely because large-scale raw ADS-B data had previously been accessible only to "a few closed industrial and governmental groups"¹Source 1 · Schäfer, Strohmeier, Lenders, Martinovic, Wilhelm — ACM/IEEE IPSN, 2014Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Researchcs.ox.ac.uk.

That makes open aviation data one of the richest public sensor feeds in existence. For a civil safety team it supports three high-value use cases: a situational air picture (who is flying where, with what confidence), anomaly alerting (emergency transponder codes, unusual altitude profiles, geofence entries), and GNSS interference mapping (where aircraft report degraded satellite-navigation quality). The third has moved from research curiosity to operational concern: EASA reports that jamming and spoofing have increased since February 2022, particularly around conflict zones, and keeps a Safety Information Bulletin on the topic, now in its fourth revision⁹Source 9 · European Union Aviation Safety Agency (EASA), 2026Global Navigation Satellite System outages and alterations (SIB 2022-02R4)easa.europa.eu. IATA data cited by EASA shows GPS signal-loss events rising 220% between 2021 and 2024¹⁰Source 10 · EASA press release, 2025EASA and IATA publish comprehensive plan to mitigate GNSS interferenceeasa.europa.eu.

The same openness creates the limits. ADS-B positions are self-reported and the protocol has an "inherent lack of security measures"⁶Source 6 · Strohmeier, Lenders, Martinovic — IEEE Communications Surveys & Tutorials (arXiv preprint), 2015On the Security of the Automatic Dependent Surveillance-Broadcast Protocolarxiv.org. Coverage depends on where receivers happen to be installed. Some aircraft, including state aircraft on sensitive missions, are legitimately authorized not to transmit²Source 2 · Cornell Law School Legal Information Institute (US Code of Federal Regulations)14 CFR § 91.225 — ADS-B Out equipment and uselaw.cornell.edu. And individual flights belong to people and companies with privacy interests that regulators recognize¹³Source 13 · Federal Aviation AdministrationADS-B Privacy (Privacy ICAO Address and LADD)faa.gov¹⁴Source 14 · EUR-Lex, Official Journal of the European Union, 2016Regulation (EU) 2016/679 (General Data Protection Regulation), Articles 4 and 5eur-lex.europa.eu. Our recommendations, developed in the chapters below, are:

1. 1Treat the air picture as an estimate with explicit confidence, not as ground truth. Carry position source, receiver count and navigation-quality fields through every model and show them in every view.
2. 2Model coverage as data. Compute where your receivers can hear, per altitude band and per hour, and render gaps as gaps.
3. 3Aggregate before you alert. A single aircraft with a degraded NACp is noise; many aircraft degrading in the same cell and time window is a signal.
4. 4Keep rules few, debounced and explained, with an owner and a review cadence for each, and measure alert precision as a first-class metric.
5. 5Govern the data like personal data: purpose-limited, minimized, retained only as long as necessary, with respect for privacy programs and data licenses.
6. 6Never use open data as the sole basis for an operational decision. Its job is to direct attention and corroborate, not to replace certified surveillance or official reporting channels.

§02What is open aviation data, and where does it come from?

"Open aviation data" in this paper means signals transmitted by aircraft for cooperative surveillance that can be received with commodity hardware, plus the public reference data needed to interpret them. Three techniques matter: ADS-B, Mode S, and multilateration (MLAT). They are often conflated in flight-tracking apps, but their provenance and trustworthiness differ, and a serious pipeline keeps them apart.

ADS-B: the aircraft tells you where it is

In ADS-B, an aircraft determines its own position, normally from GNSS, and broadcasts it periodically along with velocity, identification, status and an uncertainty level; position is typically sent twice per second, while status and intent are event-driven¹Source 1 · Schäfer, Strohmeier, Lenders, Martinovic, Wilhelm — ACM/IEEE IPSN, 2014Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Researchcs.ox.ac.uk. On the 1090 MHz Extended Squitter link, each frame is 112 bits: a 5-bit downlink format, a 24-bit ICAO aircraft address, a 56-bit payload and 24 parity bits⁴Source 4 · Junzi Sun, TU Delft (mode-s.org)The 1090 Megahertz Riddle (2nd ed.) — ADS-B basicsmode-s.org. Downlink Format 17 is used by Mode S transponders; Downlink Format 18 is used by non-transponder ADS-B equipment and TIS-B ground rebroadcasts⁴Source 4 · Junzi Sun, TU Delft (mode-s.org)The 1090 Megahertz Riddle (2nd ed.) — ADS-B basicsmode-s.org, a distinction worth preserving because a DF18 TIS-B target is a ground system relaying radar-derived data, not the aircraft speaking.

There are two ADS-B links in practice. 1090ES is used worldwide. In the United States, aircraft may alternatively use UAT on 978 MHz below 18,000 ft MSL, while operations in Class A airspace require equipment meeting the 1090 MHz standard (RTCA DO-260B, or the newer DO-260C)²Source 2 · Cornell Law School Legal Information Institute (US Code of Federal Regulations)14 CFR § 91.225 — ADS-B Out equipment and uselaw.cornell.edu. For an analyst this means a 1090-only receiver in the United States will systematically miss part of the general aviation fleet at lower altitudes, a coverage bias that has nothing to do with geography.

Mode S: interrogated replies with less context

Mode S is the selective-addressing secondary surveillance system on which 1090ES is built. Its uplink uses 1030 MHz and its downlink 1090 MHz¹Source 1 · Schäfer, Strohmeier, Lenders, Martinovic, Wilhelm — ACM/IEEE IPSN, 2014Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Researchcs.ox.ac.uk. Ground radars interrogate aircraft and receive replies containing identity, altitude and, depending on the interrogation, other parameters. A passive receiver hears these replies too, but a reply to someone else's radar carries no position: you learn that an address exists and what altitude it reports, not where it is. Mode S-only aircraft therefore appear in raw message counts but not on maps unless something else locates them.

Multilateration: locating the transmitter, not trusting it

Multilateration computes an aircraft's position from the time difference of arrival (TDOA) of the same transmission at several synchronized receivers. Each pair of receivers constrains the aircraft to a hyperboloid; with four receivers a 3D position can be estimated, and with three a 2D position if altitude is known from another source⁵Source 5 · NLR for EUROCONTROL, 2005Wide Area Multilateration — Report on EATMP TRS 131/04eurocontrol.int. Because it uses transmissions aircraft already make, MLAT requires no airborne changes⁵Source 5 · NLR for EUROCONTROL, 2005Wide Area Multilateration — Report on EATMP TRS 131/04eurocontrol.int. Its great virtue for open-data work is independence: it locates the transmitter physically instead of believing the position the transmitter reports, which is why the security literature lists it among plausibility-checking techniques⁶Source 6 · Strohmeier, Lenders, Martinovic — IEEE Communications Surveys & Tutorials (arXiv preprint), 2015On the Security of the Automatic Dependent Surveillance-Broadcast Protocolarxiv.org. Its cost is density: you need several well-synchronized receivers with line of sight to the same aircraft.

§03How do open receiver networks work, and where are they blind?

A receiver is an antenna, a radio front end and a decoder. A network is many of them streaming decoded messages, with timestamps and receiver identifiers, to a central service. The OpenSky Network began with 11 sensors hosted by volunteers in Central Europe; its founding paper reported a sensing range of about 720,000 km² and capture of more than 30% of European commercial air traffic after roughly two years¹Source 1 · Schäfer, Strohmeier, Lenders, Martinovic, Wilhelm — ACM/IEEE IPSN, 2014Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Researchcs.ox.ac.uk. The network stores, alongside each message, the receiver's identifier, a high-resolution reception timestamp and the raw frame¹Source 1 · Schäfer, Strohmeier, Lenders, Martinovic, Wilhelm — ACM/IEEE IPSN, 2014Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Researchcs.ox.ac.uk, which is what makes later research such as MLAT and interference studies possible.

Four kinds of blindness

• Geographic blindness. Receivers are ground-based and line-of-sight. Oceans, deserts, mountains and conflict areas are thin or empty. A 2022 interference study explicitly notes that large parts of its study area over the Black Sea had limited or no ADS-B coverage¹¹Source 11 · Figuet, Waltert, Felux, Olive — Engineering Proceedings 28(1), 12 (OpenSky Symposium), 2022GNSS Jamming and Its Effect on Air Traffic in Eastern Europedigitalcollection.zhaw.ch.
• Altitude blindness. Line of sight shrinks with altitude: a receiver that hears airliners at cruise 300 km away may not hear a helicopter 30 km away behind a ridge. Low-level traffic is systematically under-observed.
• Equipment blindness. Aircraft that are not required to carry ADS-B Out, or that use UAT where you only receive 1090 MHz, are absent or present only as Mode S replies.
• Authorized silence. US rules allow the FAA to authorize aircraft on sensitive government missions not to transmit, and ATC to direct transmissions off²Source 2 · Cornell Law School Legal Information Institute (US Code of Federal Regulations)14 CFR § 91.225 — ADS-B Out equipment and uselaw.cornell.edu. Absence of a track is therefore never evidence of absence of an aircraft.

Network behavior also changes over time. Volunteers move, antennas fail, and new sensors appear. A coverage model must be recomputed regularly, and any trend analysis (more emergencies this year, more interference this month) must normalize by coverage, ideally by counting observed flights per cell as the denominator. The 2014 OpenSky paper illustrates a related subtlety: it splits messages into flights using a ten-minute silence threshold, a pragmatic tradeoff that can wrongly merge a quick turnaround or split a flight that leaves and re-enters coverage¹Source 1 · Schäfer, Strohmeier, Lenders, Martinovic, Wilhelm — ACM/IEEE IPSN, 2014Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Researchcs.ox.ac.uk. Any derived "flight" is a modeling choice, and the choice should be documented.

§04Data quality: which fields can you trust?

Trust in open aviation data is field-specific. Some fields are transmitted with integrity metadata; others are typed in by crews; some are inferred by the network. The table below is the trust map we use when modeling state vectors in Kimo.

Three quality problems deserve special attention. First, not every position is GNSS-derived: the OpenSky founders observed that some aircraft broadcasting positions are not even equipped with GNSS and use less accurate means, which "can lead to large errors"¹Source 1 · Schäfer, Strohmeier, Lenders, Martinovic, Wilhelm — ACM/IEEE IPSN, 2014Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Researchcs.ox.ac.uk. Second, quality indicators were not always populated: in 2014 the same authors found the navigation-accuracy field set to "unknown" on most transponders¹Source 1 · Schäfer, Strohmeier, Lenders, Martinovic, Wilhelm — ACM/IEEE IPSN, 2014Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Researchcs.ox.ac.uk. The US performance rule has since raised the floor (NACp better than 0.05 NM and NIC better than 0.2 NM for compliant aircraft³Source 3 · Cornell Law School Legal Information Institute (US Code of Federal Regulations)14 CFR § 91.227 — ADS-B Out equipment performance requirementslaw.cornell.edu), but international and older fleets vary. Third, the address is not an identity: the 24-bit ICAO address uniquely identifies a transponder but can be reprogrammed⁴Source 4 · Junzi Sun, TU Delft (mode-s.org)The 1090 Megahertz Riddle (2nd ed.) — ADS-B basicsmode-s.org, and privacy programs deliberately issue alternate addresses¹³Source 13 · Federal Aviation AdministrationADS-B Privacy (Privacy ICAO Address and LADD)faa.gov.

Finally, ADS-B offers no cryptographic authentication. The academic literature has described this lack of security measures for over a decade⁶Source 6 · Strohmeier, Lenders, Martinovic — IEEE Communications Surveys & Tutorials (arXiv preprint), 2015On the Security of the Automatic Dependent Surveillance-Broadcast Protocolarxiv.org. In practice it means every pipeline should run plausibility checks: kinematic consistency (does the next position follow from the last one at the reported speed?), multi-receiver consistency (do several receivers hear the same target?), and MLAT cross-checks where receiver density allows. Tracks failing those checks are not necessarily malicious, since decoding errors and multipath are far more common, but they should be flagged as low confidence rather than silently drawn.

§05Emergency squawks and anomaly signals

A squawk code is the four-digit Mode A code selected on the transponder. Three codes carry special meaning in civil aviation, and automated ATC facilities are built to flag them. The FAA's Aeronautical Information Manual instructs pilots to select 7700 for emergencies ("SQUAWK MAYDAY")⁷Source 7 · Federal Aviation AdministrationAeronautical Information Manual, Chapter 4 Section 1 (4-1-20 Transponder and ADS-B Out Operation; 4-1-21 phraseology)faa.gov, 7600 after losing two-way radio capability⁸Source 8 · Federal Aviation AdministrationAeronautical Information Manual, Chapter 6 Section 4 (6-4-2 Transponder Operation During Two-way Communications Failure)faa.gov, and describes 7500 as the hijack code⁷Source 7 · Federal Aviation AdministrationAeronautical Information Manual, Chapter 4 Section 1 (4-1-20 Transponder and ADS-B Out Operation; 4-1-21 phraseology)faa.gov. It also warns pilots to avoid inadvertently selecting 7500, 7600 or 7700 when changing codes, because doing so causes "momentary false alarms at automated ground facilities"; codes in the 7600–7677 and 7700–7777 series trigger special indicators, while only the non-discrete 7500 is decoded as a hijack⁷Source 7 · Federal Aviation AdministrationAeronautical Information Manual, Chapter 4 Section 1 (4-1-20 Transponder and ADS-B Out Operation; 4-1-21 phraseology)faa.gov.

The AIM's warning about momentary false alarms translates directly into alert design: require the code to persist across several messages or seconds before alerting, and suppress alerts for a code seen in a single frame from a single receiver. Note too that open-data emergency alerts are observations, not reports. Air traffic control has the radio, the flight plan and the crew; an open-data team has a broadcast. The right response to a 7700 in Kimo is to bring context together quickly for someone with a legitimate interest (an airport operator, a civil protection duty officer, a newsroom verifying a story) and to avoid amplifying anything about a 7500 event publicly.

Other anomaly signals worth modeling

• Unusual altitude profiles: rapid descents outside approach phases, prolonged level-offs at unusual altitudes, or repeated climbs and descents.
• Holding and loitering: sustained turning in a confined area, which is routine near busy airports and noteworthy elsewhere.
• Geofence entries: civil safety zones such as temporary flight restrictions around disasters or major events, defined from official notices.
• Data anomalies: sudden position jumps, quality-field collapse, or an address appearing in two places at once. These are often reception artifacts, and they are also the first symptoms of navigation interference.

Each of these needs a baseline. "Unusual" is relative to the route, the airport and the hour. The companion guide Airspace alerting rules that analysts trust gives concrete rule definitions, thresholds and suppression windows.

§06How can aircraft data reveal GNSS interference?

GNSS interference is any radio-frequency effect that prevents a satellite-navigation receiver from computing a correct position. EASA distinguishes jamming, which blocks the signal, from spoofing, which feeds counterfeit signals to deceive receivers, and lists symptoms such as position discrepancies, time shifts and spurious terrain-warning alerts⁹Source 9 · European Union Aviation Safety Agency (EASA), 2026Global Navigation Satellite System outages and alterations (SIB 2022-02R4)easa.europa.eu. Its bulletin treats the issue as a safety matter for operators and air navigation service providers, recommending that ANSPs collect and communicate information on GNSS degradation and keep conventional navigation aids available⁹Source 9 · European Union Aviation Safety Agency (EASA), 2026Global Navigation Satellite System outages and alterations (SIB 2022-02R4)easa.europa.eu.

Aircraft are, in effect, a dense network of GNSS receivers flying through the sky and reporting how confident they are in their position. When interference denies GNSS, the navigation-quality fields they broadcast degrade. Researchers have shown that NACp drops from values above 8 to 0 when aircraft are affected by radio-frequency interference, and recovers when they leave the area¹¹Source 11 · Figuet, Waltert, Felux, Olive — Engineering Proceedings 28(1), 12 (OpenSky Symposium), 2022GNSS Jamming and Its Effect on Air Traffic in Eastern Europedigitalcollection.zhaw.ch. NIC is carried in every airborne position message (every 0.4–0.6 s) while NACp travels in the less frequent operational status message (every 2.4–2.6 s), which is why some studies prefer NIC as a per-position proxy for GNSS reception quality¹²Source 12 · Liu, Lo, Blanch, Chen, Walter — NAVIGATION: Journal of the Institute of Navigation 72(3), 2025Locating GNSS Interference Sources using ADS-B with Non-linear Least Squaresnavi.ion.org.

Figuet and colleagues used OpenSky data to study Eastern Europe from February to August 2022. To avoid flagging aircraft that never report good accuracy (for example aircraft without GNSS receivers), they labelled a flight as affected only if it reported NACp = 0 for more than 60 seconds cumulatively and NACp above 7 for more than 60 seconds¹¹Source 11 · Figuet, Waltert, Felux, Olive — Engineering Proceedings 28(1), 12 (OpenSky Symposium), 2022GNSS Jamming and Its Effect on Air Traffic in Eastern Europedigitalcollection.zhaw.ch. On 5 June 2022, more than 1,325 aircraft in their study area were affected, about 60% of the traffic, and affected flights spent on average less than 11 minutes with NACp at 0, though some were affected for over an hour¹¹Source 11 · Figuet, Waltert, Felux, Olive — Engineering Proceedings 28(1), 12 (OpenSky Symposium), 2022GNSS Jamming and Its Effect on Air Traffic in Eastern Europedigitalcollection.zhaw.ch. They also found that low-flying aircraft were less affected than those at higher altitudes, consistent with line-of-sight propagation¹¹Source 11 · Figuet, Waltert, Felux, Olive — Engineering Proceedings 28(1), 12 (OpenSky Symposium), 2022GNSS Jamming and Its Effect on Air Traffic in Eastern Europedigitalcollection.zhaw.ch.

The method has well-documented limits, and they should be printed on every map. Aircraft with tightly integrated GNSS and inertial navigation may report degradation late, distorting both the area and the duration of an event; flights already at NACp 0 when they enter an area cannot be detected; and coverage over water or in conflict areas may be thin or absent¹¹Source 11 · Figuet, Waltert, Felux, Olive — Engineering Proceedings 28(1), 12 (OpenSky Symposium), 2022GNSS Jamming and Its Effect on Air Traffic in Eastern Europedigitalcollection.zhaw.ch. Localizing the source of interference from aircraft reports is an active research area with significant assumptions (one source, static, unobstructed propagation, adequate receiver coverage)¹²Source 12 · Liu, Lo, Blanch, Chen, Walter — NAVIGATION: Journal of the Institute of Navigation 72(3), 2025Locating GNSS Interference Sources using ADS-B with Non-linear Least Squaresnavi.ion.org. We recommend that civil teams treat interference maps as indicators of where aircraft experienced degraded navigation, and leave source attribution to spectrum authorities and official reporting channels.

The step-by-step method, including thresholds, minimum-sample rules and SQL, is in the guide Detect GNSS interference from aircraft data, and a ready-made model is available as the GNSS interference monitor template.

§07Fusing tracks with weather, NOTAMs and reference data

A track on its own answers "where". Decision support needs "why" and "so what". Data fusion means joining independent sources on shared keys (time, place, identity) so that each corroborates or qualifies the others. For airspace awareness, five context layers do most of the work.

Two fusion patterns are especially valuable. Interference corroboration: when an interference cell appears, check whether an official notice already covers it. The EASA–IATA plan calls for standardized NOTAM Q-codes for GNSS interference and for timely sharing of interference event data, including between civil and military authorities¹⁰Source 10 · EASA press release, 2025EASA and IATA publish comprehensive plan to mitigate GNSS interferenceeasa.europa.eu. A cell that matches a published notice is confirmation; a cell that does not is a candidate for reporting through the proper channel. Emergency context: when a 7700 appears, automatically attach the latest weather at nearby aerodromes, any active restrictions, and the aircraft's altitude profile for the last 15 minutes. The analyst should not have to open five tabs.

In Kimo, each layer is a source in the same semantic model, so the join logic is written once and reused by the map, by alerts and by Ask Kimo. Combining aircraft data with maritime AIS works the same way, and the dark vessels article describes the gap-detection approach we reuse for tracks.

§08Designing alerts analysts will not mute

Airspace data produces an endless supply of "interesting" events: every go-around, every hold, every reception glitch. An alerting layer that surfaces all of them will be muted within a week. The aim is the opposite: a small number of alerts, each of which an analyst would agree deserved interruption. We use five design rules.

1. 01
   Step 1:

   Alert on outcomes, enrich with causes

   Page on the event a human must act on (a persistent 7700, a cell crossing an interference threshold, a track entering an active restricted area). Attach supporting signals as context instead of alerting on each separately.

2. 02
   Step 2:

   Debounce everything

   Require persistence across messages, receivers and time. The AIM documents that emergency codes can be selected momentarily by accident⁷Source 7 · Federal Aviation AdministrationAeronautical Information Manual, Chapter 4 Section 1 (4-1-20 Transponder and ADS-B Out Operation; 4-1-21 phraseology)faa.gov; a single-frame squawk should never page anyone.

3. 03
   Step 3:

   Normalize by coverage and baseline

   Compare against the same cell, hour and day-of-week baseline and only evaluate where coverage is adequate. "More events" in a cell that gained a receiver is not news.

4. 04
   Step 4:

   Group and suppress

   Group alerts by incident (one aircraft, one cell cluster) and suppress repeats within a window. An interference episode is one alert that updates, not two hundred.

5. 05
   Step 5:

   Measure precision and own each rule

   Every rule has an owner, a written purpose and a monthly review of how many alerts were useful. Retire rules that are not.

Kimo expresses alert rules as versioned YAML next to the data model, so a threshold change is reviewed like code and visible in the activity log. Examples for emergency squawks, loitering, altitude anomalies, geofences and interference cells are in the alerting guide and the alerts documentation.

§09Privacy, licensing and the law

Signals being receivable does not make every use of them appropriate. Three bodies of rules apply to most open-data airspace programs: aviation privacy programs, data-protection law, and the license terms of the data you consume.

Aviation privacy programs

The FAA runs two complementary programs. LADD (Limiting Aircraft Data Displayed) blocks aircraft data provided through the FAA's own data feed; it does not affect what the aircraft broadcasts¹³Source 13 · Federal Aviation AdministrationADS-B Privacy (Privacy ICAO Address and LADD)faa.gov. The Privacy ICAO Address (PIA) program lets eligible owners request an alternate, temporary ICAO address that is not assigned to them in the civil aircraft registry, specifically to limit how easily an aircraft can be identified by inexpensive receivers¹³Source 13 · Federal Aviation AdministrationADS-B Privacy (Privacy ICAO Address and LADD)faa.gov. A responsible pipeline honors both: it does not attempt to re-identify PIA addresses, and it applies LADD-style suppression in any public-facing output.

Data protection

Under the GDPR, personal data is "any information relating to an identified or identifiable natural person"¹⁴Source 14 · EUR-Lex, Official Journal of the European Union, 2016Regulation (EU) 2016/679 (General Data Protection Regulation), Articles 4 and 5eur-lex.europa.eu. A flight track of a privately owned aircraft can relate to an identifiable person, directly or by combination with registry data. Where GDPR applies, the core principles bite: data must be collected for "specified, explicit and legitimate purposes", be "limited to what is necessary", and be kept in identifiable form "no longer than is necessary"¹⁴Source 14 · EUR-Lex, Official Journal of the European Union, 2016Regulation (EU) 2016/679 (General Data Protection Regulation), Articles 4 and 5eur-lex.europa.eu. Practically: define the purpose of each dataset (for example "regional GNSS interference monitoring"), aggregate where individual tracks are not needed (interference maps work on cells, not aircraft), and set retention per table. This is general information, not legal advice; involve your data-protection officer.

Data licensing

Open does not mean unlicensed. The OpenSky Network, for instance, grants access for non-profit research and education; any use by a for-profit or commercial entity, including government and military contractors, requires a written license, and operational use of its REST API, even internal, requires a prior written agreement¹⁵Source 15 · OpenSky NetworkGeneral Terms of Use & Data License Agreementopensky-network.org. Publications using its data must cite the founding paper¹⁵Source 15 · OpenSky NetworkGeneral Terms of Use & Data License Agreementopensky-network.org. Kimo's OpenSky connector asks you to confirm your license tier at setup, and we recommend recording the license basis for every source in the data catalog.

§10Ethics and the limits of open data

Open aviation data sits squarely in the OSINT tradition: information anyone can lawfully collect, made powerful by aggregation. Aggregation is also where the ethical risk lives. We think four commitments separate responsible programs from irresponsible ones.

1. 1Civil purpose. Use the data for safety, research, crisis response, environmental analysis and verification. Do not use it to target, follow or harass people or to locate specific military assets.
2. 2Never the sole basis for a decision. Open data is incomplete (coverage), unauthenticated (protocol) and sometimes wrong (equipment). It should prompt questions to authoritative sources, not replace them. Interference indications go to official reporting channels; emergencies are handled by ATC.
3. 3Honest uncertainty. Every view shows coverage, source and quality. Every chart that is not measured says so. Every alert explains which rule fired and why.
4. 4Proportionate retention and access. Keep the least detail for the shortest time, and restrict sensitive alert types to people who need them.

Much has changed since 2014: equipage mandates such as the US rule have been in force since 2020²Source 2 · Cornell Law School Legal Information Institute (US Code of Federal Regulations)14 CFR § 91.225 — ADS-B Out equipment and uselaw.cornell.edu, and quality floors are now regulated³Source 3 · Cornell Law School Legal Information Institute (US Code of Federal Regulations)14 CFR § 91.227 — ADS-B Out equipment performance requirementslaw.cornell.edu. What has not changed is the distinction between certified surveillance, operated with redundancy and safety assurance by air navigation service providers, and an open-data pipeline built from volunteer receivers. The two can inform each other; they are not interchangeable.

§11A reference architecture in Kimo

The architecture below is how we deploy Kimo Defense Intelligence for civil airspace programs. It is deliberately boring: a small number of models with clear grain, reused everywhere.

Data residency is often the deciding constraint for civil-protection and research teams. With Kimo Bridge, the bridge runs on your own server next to your receiver database and opens an outbound-only, mutually authenticated tunnel; Kimo queries through it and nothing is stored on Kimo's side unless you enable short-lived caches. Teams that need fully disconnected operation run Kimo on-premise, as described in Air-gapped by design. Either way, the same models and rules apply, and the Airspace watch template provides a starting workspace with simulated data so you can evaluate the design before connecting anything real.

123456789101112131415model: nav_quality_cells
source: state_vectors
grain: [h3_cell, window_start]
window: 1h
filters:
  - position_source = 'adsb'
  - nacp is not null
measures:
  aircraft_observed: count_distinct(icao24)
  aircraft_degraded: count_distinct(icao24) where degraded_flag
  degraded_share: aircraft_degraded / nullif(aircraft_observed, 0)
gating:
  min_aircraft_observed: 8
  require_coverage: true
retention: 400d   # aggregates only; raw state vectors expire after 30d

§12A maturity model for open-data airspace awareness

Teams rarely build all of this at once. The maturity model below is a way to sequence the work and to be honest about what a given stage can support.

Note that the last column never becomes empty. Even a fully governed program does not replace certified air traffic surveillance or official interference reporting; it makes a civil team better informed and faster to ask the right question of the right authority.

§13Methodology and limitations of this paper

This paper synthesizes primary regulatory text (14 CFR 91.225 and 91.227, the FAA Aeronautical Information Manual), safety authority publications (EASA's GNSS outage bulletin page and the EASA–IATA mitigation plan), the FAA's aviation privacy program pages, the GDPR, the OpenSky Network's founding paper and license terms, a EUROCONTROL-commissioned multilateration study, and peer-reviewed research on interference detection from ADS-B. All sources were accessed in October 2026 and are listed below. Where we state Kimo design recommendations, they reflect our engineering practice, not external findings.

• Regulatory scope. Equipage rules cited are US rules; other regions differ. Check the rules where you operate.
• Evolving guidance. EASA's bulletin has been revised several times; consult the current revision rather than any summary, including this one.
• Research generalization. Interference studies cited cover specific regions and periods; thresholds that worked there may need adjustment elsewhere.
• Illustrative figures. Charts labelled "Illustrative data" are simulated and should not be read as measurements.
• Not legal advice. The privacy and licensing chapter is general information for planning.

§14Conclusion

Open aviation data has made something remarkable possible: a civil team can see most cooperative air traffic, recognize emergencies within seconds, and map where aircraft struggle to navigate, all without a radar. The craft lies in what surrounds the dots: coverage modeled as data, quality fields carried everywhere, alerts that respect an analyst's attention, and governance that respects the people behind the tracks. Build in that order and open data becomes a trustworthy instrument for safety and research. Skip it and it becomes a confident-looking map of things that may not be true.

If you want to see the patterns in this paper running end to end, open the simulated Airspace view, start from the Airspace watch template, or read the hands-on guides on ingesting ADS-B feeds, detecting GNSS interference and alerting rules.