Context-adaptive detection of insider attacks in VANET information dissemination schemes

Information dissemination is one of the most-discussed applications for vehicular ad hoc networks (VANETs) and other ad hoc networks. To provide dependability for applications, information dissemination must be resilient against different kinds of attacks. Especially insider attackers, which may create valid messages that cannot easily be detected using cryptographic signatures alone, pose a viable threat to information dependability. Many proposals in existing work offer solutions to detect individual attack patterns using data consistency checks and other means. We propose a generic framework that can integrate a wide range of existing detection mechanisms, allows to combine their outputs to improve attack detection, and enables mechanism adaptation based on current attack likelihood. We employ subjective logic opinions, which enable flexible security mechanism output representation, and which we extend to support continuing operation in dynamic networks, such as VANETs. Simulation results show that our framework improves detection accuracy compared to applying individual mechanisms.