Cybersecurity for Mixed-Autonomy Traffic: Detection and Localization of Compromised Vehicles
Automated vehicles provide new opportunities for malicious actors to compromise vehicle security. While obvious hacks that cause crashes may be easy to identify and isolate, other vehicle compromises may be more difficult to identify, especially if the hack changes vehicle driving behavior are more subtle. Such a hack could be introduced to automated or partially automated vehicles such as adaptive cruise control vehicles via a malicious software update and go undetected. However, even subtle changes to driving behavior may have widespread disruption to the transportation network by seeding new traffic jams causing delay and excess fuel consumption and emissions. For example, if such a hack were released on all vehicles of a specific make and model, even just slightly more aggressive driving could cause a network-wide increase in delays. In this work, the study team proposes to use trajectory anomaly detection to identify compromised vehicles based on their driving behavior and the collective traffic dynamics. Specifically, the team proposes to use car following models to simulate traffic flow both of typical mixed autonomy traffic as well as traffic where some of the automated vehicles have been compromised. Then, using this synthetic traffic data, the study team will use anomaly detection in the form of neural networks and autoencoders to identify atypical traffic conditions where traffic state evolution does not follow typical traffic dynamics. This work will couple traffic flow theory and emerging methods in deep neural networks to conduct the first step in a larger effort to secure transportation infrastructure against cyber threats.
Language
- English
Project
- Status: Active
- Funding: $43973
-
Contract Numbers:
CTS #2023019
-
Sponsor Organizations:
Center for Transportation Studies, University of Minnesota
University Office Plaza, Suite 440
2221 University Avenue SE
Minneapolis, MN United States 55414 -
Principal Investigators:
Stern, Raphael
- Start Date: 20220701
- Expected Completion Date: 20230630
- Actual Completion Date: 0
Subject/Index Terms
- TRT Terms: Autonomous vehicles; Computer security; Detection and identification systems; Manual control; Motor vehicles; Neural networks; Traffic flow theory; Vehicle mix
- Subject Areas: Data and Information Technology; Highways; Operations and Traffic Management; Security and Emergencies; Vehicles and Equipment;
Filing Info
- Accession Number: 01855094
- Record Type: Research project
- Source Agency: University of Minnesota, Minneapolis
- Contract Numbers: CTS #2023019
- Files: RIP
- Created Date: Aug 19 2022 11:48AM