OpenSky Report 2020: Analysing in-flight emergencies using big data Xavier Olive ¶‖ , Axel Tanner ∗∗ , Martin Strohmeier ¶†§ , Matthias Sch¨ afer ¶‡∗ , Metin Feridun § , Allan Tart ¶ , Ivan Martinovic ¶† , Vincent Lenders ¶§ ¶ OpenSky Network, Switzerland lastname@opensky-network.org ‖ ONERA, Universit´ e de Toulouse, France xavier.olive@onera.fr ∗∗ IBM Research, Switzerland axs@zurich.ibm.com ∗ TU Kaiserslautern, Germany schaefer@cs.uni-kl.de † University of Oxford, UK firstname.lastname@cs.ox.ac.uk ‡ SeRo Systems, Germany schaefer@sero-systems.de § armasuisse, Switzerland firstname.lastname@armasuisse.ch Abstract—Transponder codes are four octal digit numbers transmitted by an aircraft transponder in response to a secondary surveillance radar interrogation. These discrete transponder codes (also known as squawk codes) help with the clear labelling of an affected aircraft on radar screens. Three particular squawk codes are associated with specific situations: 7500 for hijacking, 7600 for radio failure and 7700 for general emergencies, often related to medical or technical issues. In this paper, we analyse more than 800 trajectories received by the OpenSky Network over a two-year period as they were broadcasting the 7700 emergency code. Background information about the reason of these emergencies is taken from social networks and other crowdsourced information sources on the Internet. We provide an overview of various reasons for in- flight emergencies, typical trajectory patterns and communica- tion strategies by airlines. Based on our semi-labelled dataset of trajectories, we also train models able to suggest possible explanations for trajectories when no information is available. I. I NTRODUCTION A transponder is a key electronic on-board device which helps identifying aircraft on air traffic control (ATC) radars. Traditionally, transponders produce a response on the 1090 MHz frequency after receiving a radio-frequency inter- rogation on 1030 MHz from Secondary Surveillance Radar (SSR) systems (both from the ground and other aircraft). In particular, they support the transponder Modes A, C and S, traffic alert and collision avoidance systems (TCAS) [1], and the novel Automatic Dependent Surveillance – Broadcast (ADS–B), which does not require interrogations. Air traffic controllers use the term squawk to assign air- craft a transponder code which is used to identify an air- craft uniquely in a distinct flight information region (FIR). The most elementary information transmitted by transpon- ders (Mode A/C) include information about pressure altitude (Downlink Format 4) and identification (or squawk code, Downlink Format 5). Squawk codes are made of four octal digits; the traditional dials on a transponder read from zero to seven. In normal use, squawk codes are assigned to an aircraft by ATC and subsequently applied by the pilot. Beyond this procedure, there are conventions for squawk codes, which can be selected by aircraft if and when the situation requires or allows it, without permission from ATC. Three such emergency codes are applicable worldwide: 7500 is reserved for hijacking situations and may not be used during training as it triggers a very strict security protocol; 7600 reports a radio failure to the ATC, and 7700 is reserved for general emergencies. Usually, emergencies are first reported to the ATC over the radio while the crew is assessing the situation or running through checklists (after following the old adage, “aviate, navigate, communicate”). Depending on the nature of the issue, on its gravity and on provided facilities for the airline in different airfields, pilots may choose to divert the aircraft or to continue to their final destination. When diverting, aircraft type and airline procedures may require specific manoeuvres and dumping fuel before landing. Squawking 7700 helps ATC tak- ing the emergency into account in terms of separation, priority and logistics: they may have to get Aircraft Rescue and Fire Fighting (ARFF) ready on the runway if necessary. Finally, there are regular occurrences where transponder squawk codes have been set mistakenly and/or only for a short time. It is important to detect and filter such occurrences. The wide variety of aircraft emergency situations include technical issues, e.g. landing gear failing to retract, pressuri- sation issues, cracked windshield; fuel issues, e.g. after holding or several failed attempts to land; and passenger issues, e.g. serious medical trouble or unruly passengers. Collecting and analysing such situations on a larger scale can provide two benefits: First, the insights from the use of transponder codes in real emergencies may assist in future safety analyses of emergency situations. Second, comparative knowledge about the differences between airspaces in their handling of 7700 codes could provide incentives to refine such procedures. In this paper, we present the first large-scale study about emergency squawks, their causes and consequences. In order to present a comprehensive view, we collect open-source information about worldwide emergency situations over two years (2018 and 2019). Trajectory and squawk information are extracted from the OpenSky Network [2] database and background information related to the situations is found on external sources, including social networks such as Twitter and crowdsourced information related to incidents and accidents (such as The Aviation Herald [3]).