ADS-B Message Injection Attacks Dataset

Published: 17 March 2022| Version 1 | DOI: 10.17632/6fhw732ccz.1
Hadjar Ould Slimane, Selma Benouadah, Naima Kaabouch


This dataset includes authentic ADS-B messages samples from the OpenSky Network and injected messages simulated using PyCharm. This dataset contains a total of 22,316 messages where legitimate messages are labeled with 0 and injected malicious messages labeled with 1, 2, and 3 for the path modification, ghost aircraft injection, and velocity drift attacks, respectively. This dataset contains 17 features of which 15 are extracted from ADS-B messages, the Received Signal Strength (RSS), and the Doppler shift. A supplementary file containing raw ADS-B messages of the selected receiver is also provided. This dataset can be used to train machine learning models in order to build robust ADS-B message injection attack detection techniques.



University of North Dakota


Cybersecurity, Wireless Communication, Machine Learning