Um protocolo de autenticação e detecção de ataques sybil em redes Ad Hoc veiculares com suporte ao controle de anonimato.

Abstract

The VANET (Vehicle Ad Hoc Networks) network aims at allowing the development of intelligent transportation systems, in which the vehicles are mobile nodes that can communicate with each other to share information. In order to make it possible, each vehicle may transmit two types of messages (or events): beacon messages and event-based messages. The former announces vehicle’s current position, speed, and direction to neighbors, allowing other vehicles to perceive and predict the kinematics of the vehicle. The last one are sent in order to announce sporadic events, such as accident report, road hazard condition notification, emergency vehicle approaching etc. Within this context, each message must attach enough information in order to guarantee vehicle authentication, a fundamental security requirement to any distributed system. On the other hand, if different messages in different geographical areas are linked to the same vehicle, an intruder may build a vehicle’s route profile, a potential threat to user’s privacy. To avoid such threat, it is possible to allow a vehicle to keep multiple identities at different moment, a strategy called pseudonymity. However, although such an approach provides a simple security solution to avoid vehicle monitoring, the strategy of pseudonymity allows a malicious vehicle to use two or more identities (pseudonyms) at the same time to announce the same false event, a network attack called the sybil attack. In this perspective, this thesis proposes a privacy-preserving authentication and sybil attack detection protocol called ASAP-V. To this end, we propose an authentication model based on the group signature scheme and combines such approach with a multi-level k-anonymity set architecture to detect sybil attacks, while promoting message non-repudiation and users privacy control. Through a formal analysis of the authentication protocol and the anonymity control model, as well as through a series of experimental simulations, results suggest that ASAP-V is more efficient than the state of the art approaches, mainly because it detects legitimate and malicious vehicle faster, and is resilient to false-positive and false-negative sybil detections.