Malicious Messages Detection And Exclusion Mechanisms In Vehicular Networks (vanets)

Résumé: The wide range of applications supported by Vehicular Adhoc NETworks (VANETs) make them vulnerable to various security threats. This thesis has enabled us to identify what are the main threats, adversary models, and security mechanisms associated to VANET environments. We focused mainly on inside attackers able to bypass the classical cryptography-based security mechanisms. Inspired by economic science, trust management is considered as an alternative security solution able to handle authorized and authenticated attackers. Furthermore, trust management has shown its high adaptability to VANET features (i.e, open communication medium, very high mobility, etc.). In order to handle inside attacks on VANETs, we have proposed three different trust establishment solutions: TFDD, T-VNets, and RITA. For each of these solutions, we have studied diferent attacker behaviors, as well as heterogeneous types of attacks including Denial of Service (DoS), Blackholes, Grayholes (selective forwarding), Coalition, Platooning, Badmouthing, and detection-avoidance strategies.Furthermore, all proposals are hybrid trust models able to revoke dishonest entities and lter out malicious messages, while considering different communication scenarios and applications. Regarding the frst proposal - TFDD - it is based on a modular architecture, being able to ensure a reliable unicast and multi-hop communication even under DoS attacks, and with a high ratio of attackers.Our second proposal - T-VNets - is the frst trust model able to take advantage of the messaging services provided by the ETSI ITS standard. Through continuous trafc and trust estimations, T-VNets is able to handle not only DoS attacks, but also coalition attacks, without exhausting the network resources. Concerning our third proposal - RITA - it difers from existing trust-based security solutions by also being able to handle intelligent attack behaviors. In another words, to accounts for attackers able to adjust their behaviour with full awareness of the deployed security rules. Thanks to the introduced risk estimation metric, RITA is able to face not only intelligent attackers, but also those attacks where false recommendations and fake alerts are spread. In addition to the aforementioned proposals, our contributions also include two trust-based lightweight routing and dissemination schemes for VANET environments.Simulation results evidence shows the effciency of our proposals at both ensuring high detection ratios and reduced overhead.

Publié dans la revue: