Edge Based Lightweight Authentication Architecture Using Deep Learning for Vehicular Networks

Abstract

When vehicles are connected to the Internet through vehicle-to-everything (V2X) systems, they are exposed to diverse attacks and threats through the network connections. Vehicle-hacking attacks in the road can significantly affect driver safety. However, it is difficult to detect hacking attacks because vehicles not only have high mobility and unreliable link conditions, but they also use broadcast-based wireless communication. To this end, V2X systems need a simple but a powerful authentication procedure on the road. Therefore, this paper proposes an edge based lightweight authentication architecture using a deep learning algorithm for road safety applications in vehicle networks. The proposed lightweight authentication architecture enables vehicles that are physically separated to form a vehicular cloud in which vehicle-to-vehicle communications can be secured. In addition, an edge-based cloud data center performs deep learning algorithms to detect car hacking attempts, and then delivers the detection results to a vehicular cloud. Extensive simulations demonstrate that the proposed authentication architecture significantly enhanced the security level. The proposed authentication architecture has 94.51 to 99.8% F1-score results depending on the number of vehicles in the intrusion detection system using control area network traffic.