Data-driven Modeling and Anomaly Detection of Attacked HATS

"Since automated vehicle technologies continue to advance, their dependence on Positioning, Navigation, and Timing (PNT) systems increase to ensure safe and reliable operation. This reliance introduces new cy-bersecurity challenges, as adversarial manipulation of navigation signals, particularly GPS, can distort a vehicle’s perception of its environment and position. To safeguard the operational integrity of these sys-tems, this project emphasizes the detection of navigation-related anomalies, including those caused by GPS spoofing. It also explores the role of sensor redundancy, using diverse data sources to uncover incon-sistencies in vehicle behavior and positioning. By examining how discrepancies arise across these sys-tems, the research aims to enhance situational awareness and enable prompt responses to abnormal con-ditions in highly automated transportation systems. The project's main components are: • Detection and Identification: GPS spoofing attacks have been detected using data-driven algo-rithms based on discrepancies between GPS and IMU displacements. However, IMU faults can cause false alarms, necessitating additional data from other sensors. The primary candidates are the steering wheel angle, which is included in the Honda Driving Dataset but is less accurate, and visual data from camera, which is more precise but not available in the dataset. Visual data en-hances IMU accuracy by forming a Visual-Inertial Navigation System (VINS). Since visual data is unavailable, it will be collected at an in-indoor testing facility using camera, IMU, and uncrewed ground vehicle (UGV) to develop robust and accurate spoofing detection algorithm. • Localization and Risk Assessment under GPS Spoofing Attacks: This work addresses the chal-lenge of localization in GPS-compromised environments, ensuring that HATS can accurately deter-mine their position despite spoofing attacks. The primary goal is to develop a trajectory prediction model that allows HATS to recover localization using spatial-temporal dependencies while main-taining reliable navigation. Additionally, we seek to establish a systematic framework for quantify- ing the risk and safety implications of GPS spoofing attacks. By analyzing how spoofing impacts" "vehicle stability, route deviations, and system security, the findings will guide the development of resilient navigation strategies and inform cybersecurity policies for HATS."

• Record URL:
  • https://utc.engineering.osu.edu/data-driven-modeling-and-anomaly-detection-attacked-hats

Language

• English

Project

• Status: Active
• Funding: $459,000.00
• Contract Numbers:

  69A3552348327

• Sponsor Organizations:

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590
• Managing Organizations:

  Center for Automated Vehicle Research with Multimodal Assured Navigation

  Ohio State University
  Columbus, OH  United States  43210
• Project Managers:

  Ghasemi, Hamid

• Performing Organizations:

  North Carolina A&T State University

  1601 E. Market Street
  Greensboro, NC  United States  27411
• Principal Investigators:

  Homifar, Abdollah

• Start Date: 20231030
• Expected Completion Date: 20260831
• Actual Completion Date: 0
• USDOT Program: University Transportation Centers Program

Subject/Index Terms

• TRT Terms: Autonomous vehicles; Computer security; Data analysis; Risk management; Validation
• Subject Areas: Data and Information Technology; Highways; Security and Emergencies; Vehicles and Equipment;

Filing Info

• Accession Number: 01901373
• Record Type: Research project
• Source Agency: Center for Automated Vehicle Research with Multimodal Assured Navigation
• Contract Numbers: 69A3552348327
• Files: UTC, RIP
• Created Date: Dec 4 2023 4:58PM