An Intrusion Detection System Using the XGBoost Algorithm for SDVN

Vehicular ad hoc networks (VANETs) rely on Software-Defined Networking (SDN) to enable continuous exchange of information and messages about vehicle and road conditions. This facilitates convenience for users and improves decision-making and safety. However, the communication of Electronic Control Units (ECUs) through the Control Area Network (CAN) poses security risks. The CAN is vulnerable to a range of security attacks, including Denial of Service (DoS), fuzzy attacks, and spoofing RPM, which can cause traffic congestion, fatal accidents, or disrupt the network services provided to users. To address these security challenges, we propose an Intrusion Detection System (IDS) that uses the XGBoost machine learning algorithm. Our IDS leverages a car-hacking dataset to detect traffic patterns and classify them as normal or attack patterns. Specifically, our research examines three types of attacks: DoS, fuzzy, and spoofing RPM, which are present in the car-hacking dataset. We show that our proposed IDS outperforms external infiltration systems that use KNN and LSTM-AE algorithms. By enhancing the security of SDN-based VANETs, our proposed framework contributes to safer and more reliable vehicular communication.