Sensor Integrity Verification
Vehicle Anomaly Diagnostic SystemTo provide safer and more comfortable driving experience, more and more electronic components are added to cars. However, their integration also increases the system complexity and introduces new attack surfaces, thus making it harder to identify the source/cause of anomalous vehicle behavior. To alleviate this difficulty, we have been developing a vehicle anomaly diagnostic system, called CarDog, to trace and hunt/narrow down the source(s) of anomaly. Specifically, CarDog is designed to identify the source(s) of anomaly related to vehicle maneuvers, including input-to-response consistency and powertrain operation. It constructs hierarchical detection sets (DSs) that overlap with each other. By capturing the cyber-physical characteristics in each DS, CarDog is able to cross-validate the vehicle data with multiple combinations and identify the faulty data/components even if no data can be entirely trusted.
Context-Aware Sensor Integrity Verification
While more and more driver assistance function added to modern vehicles, they also introduce new attack surfaces to the vehicle. The goal of this project is to develop a system that is able to detect and verify context information, such as road inclination, for verifying vehicle IMU readings and location information to prevent the control system or service provider from producing an incorrect decision. The system captures the context information by monitoring the behavioral change between different vehicle components. The system then compares the captured change to the road/map information to verify the data integrity or identify the occurrence of an anomaly.
Faculty
- Kang G. Shin
Graduate Students
- Chun-Yu (Daniel) Chen
Publications
-
Chun-Yu (Daniel) Chen, Kang G. Shin, and Soodeh Dadras, Context-Aware Anomaly Detection Using Vehicle Dynamics, in the 27th International Symposium on Research in Attacks, Intrusions and Defenses (RAID '24), Pauda, Italy, October 2024.
<pdf>