Securing V2X Communication: A Hybrid Deep Learning Approach for Misbehavior Detection

The emerging vehicle-to-everything communications lead to a future of Intelligent Transportation Systems. However, this also calls for set of novel mechanisms to address new vulnerabilities and security problems. In this context, misbehavior detection approaches aim to detect malicious behavior of rogue vehicle-to-everything entities and possible attacks that may originate from them. This paper proposes a hybrid deep learning approach for misbehavior detection in 6G vehicle-to-everything communication. The approach utilizes a two-stage architecture. The evaluation results show that the proposed approach achieves high accuracy in identifying misbehaving vehicles, including unknown ones. Additionally, the best performing model outperforms existing frameworks in terms of key evaluation metrics while also being lightweight for resource-constrained environments and having real-time requirements in mind.