A Framework for Malicious Node Detection in VANETs using Machine Learning Approaches

Abstract

Vehicular Ad-hoc Networks (VANETs) are crucial for enhancing road safety, enabling autonomous driving, and supporting efficient traffic management. However, the open and highly dynamic nature of VANETs makes them vulnerable to various cyber threats posed by malicious nodes, including Sybil attacks, spoofing, and denial-of-service (DoS) attacks. This paper presents a novel hybrid detection framework that combines machine learning with a trust-based model to enhance malicious node detection in VANETs. The proposed system leverages machine learning for real-time anomaly detection by analyzing communication patterns, while the trust-based model evaluates each node's behavior and reputation to strengthen the accuracy of malicious activity detection. By integrating these approaches, the framework provides robust, adaptive, and precise detection capabilities, effectively minimizing false positives and improving resilience against emerging threats.

Our framework consists of a multi-layered architecture with data collection, feature extraction, machine learning-based anomaly detection, and a trust evaluation system that assigns trust scores based on historical and real-time interactions. The decision fusion module combines outputs from both machine learning and trust-based models to make informed decisions on potential threats. The proposed hybrid detection system significantly improves VANET security, offering an efficient and reliable solution for detecting and mitigating malicious nodes in complex vehicular networks.