Abstract:The fifth-generation (5G) technology makes it widely applicable to connected vehicles. This would entail numerous transmitted data in communication networks and frequent information interactions between vehicles and other terminals, thus leading connected vehicles to be vulnerable to attacks from external communication interfaces. This paper analyzes potential security threats of 5G vehicular network and focuses on intrusion detection methods for in-vehicle networks. We choose four experiment scenarios from potential attacks for in-vehicle networks and collect real car data to compile various attack databases for the first time. In order to find appropriate methods to identify different attacks, four light-weight intrusion detection methods are presented to recognize abnormal behaviors of in-vehicle networks. Furthermore, our study undertakes the detection performance comparison between four detection methods with considering comprehensive evaluation metrics. The evaluation results provide optimal light-weight detection solution for in-vehicle networks. This paper facilitates the understanding of the advantages of the test methods in detection performance for in-vehicle networks and promotes the application of detection technology to deal with the security issues of automotive industry.

Comparative Performance Evaluation of Intrusion Detection Methods for In-Vehicle Networks