OpenSky Network

The OpenSky Network is a non-profit association in Burgdorf, Switzerland, that collects air traffic data from receivers run by volunteers, companies, universities and government bodies, and archives the raw data in a historical database used mostly by researchers.^2,4 It began in 2012 as a research project of armasuisse, the University of Kaiserslautern and the University of Oxford, and became an independent association in 2015.²

Most of the data comes from ADS-B (Automatic Dependent Surveillance-Broadcast): an aircraft works out its position and speed by GPS and broadcasts them on the 1090 MHz radio channel.^2,3 The network also records Mode S messages on the same channel, and FLARM and ADS-C data.^2,3 OpenSky keeps both the single transponder messages and tracks of the form "at time T, aircraft A was at position X and altitude Y", and works out each flight's origin and destination airports in a nightly batch from the finished day's tracks.³ It holds no schedules, delays, flight numbers or passenger counts, nothing from before 2013, and its coverage is best in Europe and the US.³

The REST API returns state vectors, an aircraft's position, altitudes, speed, heading and callsign at a moment, for the whole world or a bounding box, and flights by time, aircraft or airport for past days; a tracks endpoint is marked experimental.^3,5 Anonymous callers get only the latest states at 10-second resolution; signed-in users, who authenticate with OAuth2 client credentials, get 5-second resolution.⁵ Calls cost credits, refilled daily and counted separately for states, flights and tracks.⁵ Anonymous use gets 400 a day, a standard user 4,000, and an active feeder whose receiver is up at least 30% of the month 8,000.⁵ The full dataset is reached through a Trino database interface, granted free to research institutions, aviation authorities and government organisations, and to registered non-profits case by case; private individuals are usually pointed to the curated datasets instead.³ One holds every Monday's state vectors from 2017 to 2022 at 10-second intervals, in CSV, Avro and JSON.⁶ Others are a COVID-19 flight dataset for 2019 to 2022, an aircraft metadata database in CSV, and a one-day snapshot of all the Trino tables as of 1 March 2026.⁶

The licence covers use "solely for the purpose of non-profit research and non-profit education".⁴ Any for-profit or commercial entity, including government and military contractors, needs a written licence, and so does "operational use of the REST API in any live product, service, or automated system", whatever the user's non-profit status.⁴ Publications, web pages included, must cite Schäfer, Strohmeier, Lenders, Martinovic and Wilhelm, "Bringing up OpenSky: A large-scale ADS-B sensor network for research" (ACM/IEEE International Conference on Information Processing in Sensor Networks, April 2014), and a copy of or link to the publication must go to OpenSky.⁴ Data may not be passed to anyone outside the user's own institution, and aircraft identifiers in non-anonymised data must be de-identified before publication.⁴

More than 800 academic papers have used OpenSky data.⁷ Journalists such as Bellingcat and the Financial Times, and institutions such as the IMF, the World Bank and the UN, have had access for non-profit work ranging from the effect of COVID-19 on air traffic to investigative reporting.⁷ For a model of a place, the state vectors over a bounding box show which aircraft pass over it and at what height, and the flight records count arrivals and departures at its airports.⁵