Securing vehicular ad-hoc networks against DDoS and Sybil attacks via machine learning techniques

Vehicular Ad hoc Networks (VANETs) are important for enhancing traffic management and road safety in Intelligent Transportation System (ITS). However, their networks may be affected by Distributed Denial of Service (DDoS) and Sybil attacks. This study performs a Systematic Literature Review (SLR) of Machine Learning (ML)-based detection methods in accordance with the PRISMA guidelines. The results show that Random Forest (RF) has an accuracy of 99.7% on the VDDD dataset for DDoS detection, whereas Extreme Learning Machine (ELM) is most effective on simulated data for Sybil detection. Challenges such as imbalance in training data and few real-world applications make it important to pursue hybrid approaches and create datasets that better reflect real situations for VANET security.