Machine Learning Approach to False Alert Attack Detection in Vehicular Adhoc Networks

The expansion of Intelligent Transportation Systems (ITS) and their integration into Vehicular Adhoc Networks (VANETs) bring a number of critical safety and security concerns. Among many is the false reporting attack, in which a malicious actor sends adversarial messages to fabricate artificial traffic incidents. To ensure safety and reliability, addressing false alert attacks is particularly crucial due to the potential danger that this type of attack poses. False alerts may cause vehicles to take unnecessary evasive maneuvers to avoid non-existent hazards, which might lead to accidents and endanger the safety of drivers, passengers, and other road users. In this research, we aim to address this threat of false alert attacks in VANETs by using the VeReMiAP Dataset (a VeReMi-based dataset) as a benchmark and develop machine learning (ML) models and approaches to detect and mitigate false alert attacks in VANETs. The methodology includes a detailed analysis of the dataset, feature engineering in conjunction with plausibility, and the use of state-of-the-art ML models for detection. The research findings show that the proposed approach can effectively detect false alert attacks in VANETs. The results show that this approach is effective for detecting the attack with high accuracy and F1 score. The research also provides insights into the performance of different ML models and the importance of feature engineering in detecting false alert attacks. The research findings can be used to develop more robust and reliable security mechanisms to ensure the safety and security of road users.