No abstract

The deployment of Connected Autonomous Vehicles (CAVs) in Vehicular Ad Hoc Networks (VANETs) requires secure wireless communication in order to ensure reliable connectivity and safety. However, this wireless communication is vulnerable to a variety of cyber atacks such as spoofing or jamming attacks. In this paper, we describe an Intrusion Detection System (IDS) based on Machine Learning (ML) techniques designed to detect both spoofing and jamming attacks in a CAV environment. The IDS would reduce the risk of traffic disruption and accident caused as a result of cyber-attacks. The detection engine of the presented IDS is based on the ML algorithms Random Forest (RF), k-Nearest Neighbour (k-NN) and One-Class Support Vector Machine (OCSVM), as well as data fusion techniques in a crosslayer approach. To the best of the authors' knowledge, the proposed IDS is the first in literature that uses a cross-layer approach to detect both spoofing and jamming attacks against the communication of connected vehicles platooning. The evaluation results of the implemented IDS present a high accuracy of over 90% using training datasets containing both known and unknown attacks.