Scaling the Timing-Based Detection of Anomalies in Real-World Aircraft Trajectories

Abstract

The insecurity of ADS-B and GPS is a well-known problem in air traffic control. In this work, we introduce a statistical location verification method for crowdsourced aircraft data. We implement this method on the backend of the OpenSky Network and evaluate it on 8 months of data collected from Radarcape sensors. We thus demonstrate the effectiveness of our method in finding anomalous flight data in faster than real time. While we do not find clear evidence of widespread malicious spoofing attacks captured by OpenSky sensors, many sensor and transponder issues are unearthed. Finally, we simulate ADS-B and GNSS attacks, which were detected successfully with few false positives.