Stateful Behavior Inference and Runtime Enforcement for Vehicle Network Security

Learn about a comprehensive security framework for vehicle communication networks in this 14-minute conference presentation from VehicleSec '25. Explore how researchers from UC Santa Barbara, University of Florida, and Oceanit Laboratories address critical security concerns in software-defined vehicles by focusing on Electronic Control Units (ECUs) connected via Controller Area Network (CAN) bus systems. Discover the innovative three-part framework that includes behavior inference techniques for capturing expected vehicle behavior during specific scenarios as state machines, hierarchical state machine models for monitoring multiple scenarios, and state-machine-based runtime behavior enforcement mechanisms using ECU fingerprinting for message authentication. Examine the presented threat model, characterization of different attack types, and security analysis based on fuzzing methodologies. Review the practical implementation and evaluation results conducted on a Toyota Prius and the BeamNG.tech simulator, demonstrating real-world applicability of the proposed security solutions for modern automotive systems.