SeeD: AI-based Secure Data Exchange Framework for Connected Electric Vehicles

The functionality of Connected Electric Vehicles (CEVs) can be greatly enhanced using Software Defined Networking (SDN), but the security of the network is an important drawback to this approach. This paper presents an AI-based framework called SeeD for filtering out malicious data flow to safeguard the network against malicious data transmission, thereby ensuring secure data exchange between two CEVs. The proposed framework involves training and testing various models, including random forest, decision trees, XGBoost classifiers, K-Nearest Neighbor (KNN) classifiers, and Neural Networks. We use feature selection methods on an SDN flow dataset. Among these models, the XGBoost model, utilizing the Fischer score feature selection method, achieves the highest accuracy of 98.6%. This model is applied following the SDN layer of the system to filter out malicious data. The results demonstrate that our model is highly efficient in classifying cyber attacks, thus enhancing network security.