Security Defense of Transportation Networks Against Cyberattacks: A Physics-Informed AI Approach

Details

• Creators:
  Kim, Yongju ; Chen, Sikai ; Ahn, Soyoung (Sue) ; Chitturi, Madhav ; Noyce, David A.
• Corporate Creators:
  Center for Connected and Automated Transportation (CCAT) ; University of Wisconsin--Madison
• Corporate Contributors:
  United States. Department of Transportation. University Transportation Centers (UTC) Program ; United States. Department of Transportation. Office of the Assistant Secretary for Research and Technology
• Subject/TRT Terms:
  Adversarial Reinforcement Learning Anomaly Detection Communication Devices Computer Security Connected Vehicles Data Access Control Decision Making Detectors Evaluation Intelligent Transportation Systems Lane Changing Neural Networks Physics-guided Neural Network Simulation

  More +
• Publication/ Report Number:
  GR000034957
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States
• Edition:
  Final report (April 2024 – March 2026)
• Corporate Publisher:
  University of Michigan. Center for Connected and Automated Transportation
• Abstract:
  Connected and automated vehicles (CAVs) rely on digitally acquired traffic state information for real-time maneuver decisions, making them vulnerable to cyberattacks that manipulate perceived traffic states or communication channels. Such manipulation can subtly alter vehicle behavior and, in interactive traffic environments, may propagate disturbances beyond the attacked vehicle. The objective of this study is to develop frameworks for (i) detecting abnormal lane-changing under cyberattacks and (ii) enabling robust decision-making under adversarial conditions. For detection, a physics-guided neural network integrating a game-theoretic lane-changing model with LSTM learning is developed and evaluated using the I-24 MOTION dataset under simulated false data injection and denial-of-service attacks. Results show improved prediction accuracy and effective detection of falsified lane-changing behaviors. For robust control, a hierarchical adversarial reinforcement learning framework is proposed for discretionary lane-changing under bounded perturbations. Simulation results indicate improved traffic efficiency while maintaining safety under worst-case perturbations. These findings provide a modeling foundation for cybersecurity-aware monitoring and robust tactical decision-making in CAV systems.
• Format:
  PDF
• Funding:
  69A3552348305
• Collection(s):
  University Transportation Centers
• Main Document Checksum:
  urn:sha-512:c8a3656004f47573e90e6e44b2a420c7d2a572bc7a1cfd13312b9cbce30e90acab095fbfed4cc1d29af4ff93c6ed344f93675f860ba8ce1727c3256069c72ec8
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/89724/dot_89724_DS1.pdf
• File Type:
  [PDF - 3.69 MB ]