IntelliMD: A Hybrid Approach for Local Misbehaviour Detection in Cooperative Intelligent Transport Systems

Abstract

Cooperative Intelligent Transport Systems (C-ITS) consist of various ITS Subsystems that are vulnerable to misbehaviours with potentially significant consequences. These misbehaviours include replaying outdated data or injecting false information into C-ITS communication messages such as the Basic Safety Message, the Cooperative Awareness Message, and Vulnerable Road Users Awareness Message. Such misbehaviours can confuse the receiving entities, potentially leading to undesirable outcomes such as accidents and road hazards. In this study, we present IntelliMD, a novel hybrid approach tailored for local (in-vehicle) misbehaviour detection within C-ITS. IntelliMD leverages rule-based message checks with unsupervised machine learning techniques to effectively identify false data injection and data replay attacks. Extensive experimental evaluations, including tests conducted on a Raspberry Pi emulating an On-Board Unit, demonstrate that IntelliMD achieves impressive accuracy of 98.14% at scale while optimising memory and CPU usage. IntelliMD also outperforms the existing literature in terms of accuracy, detection time, and resource requirements.