Cooperative Adaptive Cruise Control – Part 2: Misbehaviour Detection and Mitigation
TU Braunschweig, Ulm University
This second video shows how mechanisms developed in the SecForCARs project can help to detect and mitigate such attacks. The Misbehaviour Detection System TriP uses trust values that each platoon member calculates for its predecessor in the platoon. For that, the vehicle compares the actual behaviour, which it judges based on its local sensors, and the behaviour announced in the platooning beacons. If those match closely, the vehicle slowly builds up trust in its predecessor. If they deviate too much on the other hand, the trust values rapidly drop.
At this point, a Misbehaviour Mitigation System comes into play which was also developed in SecForCARs. It uses the trust values to estimate an attack probability for the predecessor and if that probability increases beyond a first threshold, the safety gap is gradually increased with increasing attack probability. If the value further increases and exceeds a second threshold, platooning with this respective predecessor is not considered safe at all anymore and the presumably attacked vehicle splits the platoon. It then rapidly increases its safety gap to its predecessor, the presumed attacker, to a distance that is considered safe under normal, non-cooperative conditions. If it was not the last vehicle in the old platoon, it forms a new platoon with its old successors, where the formerly attacked vehicle itself is the new platoon leader.
Both Misbehavior Detection and Misbehavior Mitigation were implemented in our demonstrator, and this video demonstrates how they can effectively avoid a crash when the platoon is attacked by injecting manipulated speed data into the platooning beacons.