Machine Learning based Hybrid Technique for Cyber Attack Prediction in VANET

This paper introduces a hybrid machine learning-based framework that combines the Random Forest and the XGBoost machine learning algorithms for effective prediction of cyber-attacks in Vehicular Ad Hoc Networks (VANETs). These networks are especially sensitive to various cyber-attacks, such as denial-of-service attacks, spoofing attacks, and the Sybil attacks, due to a very dynamic topology, decentralization architecture, and real-time communication needs in VANET environments. To overcome them, the hybrid approach is suggested, as the random forest feature selection power and preliminary classification are enabled to enable the principle of a strong feature selection with the help of random forest, after which the optimization power of the XGBoost is suggested to increase the predictive ability and reduce the level of misclassification through boosting. Its methodology includes preprocessing systematic data, feature engineering, normalizing, and class imbalance to attain trustworthy model training. The experimental outcomes suggest that the combined model is better in accuracy, precision, recall, and F1-score compared to its individual counterparts, which allows the proposed integrated model to offer a scalable and efficient system of detecting and predicting cyber-attacks in real-time in intelligent transportation systems.