Primary Investigator:
Satish Ukkusuri

Project Members
Eunhan Ka; Doguhan Yeke; Z. Berkay Celik; Satish V. Ukkusuri

The emergency electronic brake light (EEBL) application in Vehicle-to-Everything (V2X) networks warns upstream vehicles of hard braking. However, recent studies show that a compromised vehicle with valid credentials can broadcast fabricated EEBL messages, triggering unnecessary emergency braking and thereby creating significant traffic safety risks. To detect such attacks, existing misbehavior detection systems apply generic plausibility checks to every basic safety message, resulting in false alarms during normal driving and missed detections when the attacker generates a kinematically self-consistent trajectory. To address this problem, we propose MIRAGE (Multi-stage Inspection and Response Against Ghost-vehicle EEBL), an event-gated detector grounded in the Intelligent Driver Model (IDM). MIRAGE activates only under EEBL conditions, validates post-event trajectory consistency using position-derived speed, detects post-attack position freezing artifacts, and bounds the ego vehicle's response via IDM even on missed detections. We implement MIRAGE within the V2X Application Spoofing Platform (VASP) and evaluate against two fake EEBL attack variants alongside three existing defenses. MIRAGE achieves an F1 score of 0.922 across five random seeds with 88.3