Detection of false position attacks in VANETs through bagging ensemble learning

Area of Science:

• Cybersecurity
• Network Security
• Intelligent Transportation Systems (ITS)

Background:

• Vehicular Ad-hoc Networks (VANETs) are crucial for vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications in ITS.
• VANETs are vulnerable to security threats, especially position falsification attacks using false Basic Safety Messages (BSMs).

Purpose of the Study:

• To propose an ensemble learning framework for detecting position falsification attacks in VANETs.
• To evaluate the performance of various classifiers, including Decision Tree (CART), Random Forest, K-Nearest Neighbors (KNN), and Multilayer Perceptron (MLP), enhanced with bagging.

Main Methods:

• Utilized the VeReMi dataset for analysis.
• Implemented a Road-Side Unit (RSU)-level detection system analyzing sequential BSMs.
• Employed ensemble learning techniques with bagging for classifier enhancement.

Main Results:

• KNN with bagging achieved 100% precision, recall, accuracy, and F1 score for Attack 1.
• KNN with bagging demonstrated near-perfect performance for complex attacks (Attack 2: 99.87% accuracy, Attack 16: 97.85% accuracy).
• Other ensemble methods showed varying effectiveness, with KNN with bagging being the most robust.

Conclusions:

• Ensemble learning techniques, particularly KNN with bagging, are highly effective for detecting sophisticated attacks in VANETs.
• The proposed framework offers a scalable, efficient, and robust solution for securing VANET communications.
• This approach significantly enhances the security of Intelligent Transportation Systems.