Detection of Position Falsification Attacks in VANETs Using Ensemble Learning

Vehicular Ad Hoc Networks (VANETs) are utilised by Cooperative Intelligent Transport Systems (C-ITS) to facilitate smooth communication between vehicles and infrastructure, markinga significant technological advancement in traffic efficiency and road safety. However, the security of VANETs is a paramount concern in C-ITS, as potential attacks could compromise passenger safety. Intrusion Detection Systems (IDS) are pivotal in safeguarding vehicular networks by identifying misbehaving vehicles. A misbehaving network could interfere with any data in the propagated packet. Fraudulent messages can be made by the attacker node or by forcing another network node to create fraudulent messages. It introduces a novel approach to enhance IDS performance in the context of position falsification attacks. Machine learning uses algorithms to extract patterns from data, whereas ensemble learning mixes many models to increase prediction accuracy. Ensemble models outperform individual machine learning algorithms in detecting the position falsification attacks because they lower noise, incorporate various inputs, reduce bias and variance, detect abnormalities effectively, and capture complex data relationships. The performance of four distinct ML methods for classification was compared that is KNN, Random Forest, Gradient Boosting and XGBoost. Finally, the XGBoost, short for Extreme Gradient Boosting out performs with high accuracy of 97% with efficient and effectiveness which is ensemble learning technique that combine the predictions of multiple weak learners. The results demonstrate that, on average, the proposed mechanism performs better than earlier approaches as a result of computational time and performance on classification criteria.