Mitigating Link-flooding Attacks in Intelligent Transportation System

Vehicular Ad hoc Network (VANET) is an important component of intelligent transportation systems. In VANET, nodes' high mobility and limited computing resources make it easy for them to be controlled by attackers to become botnets. Link flooding attack (LFA) is a new attack type that uses botnets to send legitimate low-speed traffic to flood critical links to cut off the target area, which poses new security risks for VANET. Therefore, the paper proposes an LFA mitigation scheme based on the programmable network architecture to improve security in VANETs. First, through the designed telemetry and early warning methods, the fine-grained network state can be efficiently obtained in real-time, and the link condition can be evaluated quickly. Afterward, the reroute scheduling policy for traffic is customized with the help of graph neural networks to reduce the pressure on critical links in time. The experiment shows that the scheme can rapidly mitigate LFA.