Misbehavior Detection in Connected Vehicles using BurST-ADMA Dataset

The reliability and security of information within and between connected vehicles is now a burning research issue. Connected vehicle security is of utmost importance to military, civil, and civil-military operations. In addition, the vehicular network is a mission-critical system placing the demand for real-time and accurate detection on machine learning models. In this work, machine learning models were evaluated to guide in selection and design of reliable detection schemes. The novel dataset BurST-ADMA was used for the evaluation since it is an updated dataset with possible features for false message detection in IoV. Results show that optimizing ensemble models can enhance the accuracy and reduce mis-classification costs. The proposed AdaBoost outperformed other ensemble classifier models with an accuracy of 98.92 %.