Investigating the Effects of Attack Detection for In-Vehicle Networks Based on Clock Drift of ECUs

Abstract

As external interfaces of vehicles multiply, information security of an automobile network system has become increasingly troubling. Mounting attacks have raised the attention of researchers to seek optimal solutions. Therefore, they set forth attack detection to demonstrate the vulnerability of in-vehicle networks, yet most of them focus on packet information directly. This paper comprehensively analyzes the vulnerability of in-vehicle networks and investigates a unique detection method based on clock drift of electronic control units. To investigate the applications of the proposed method further, we take attack time and attack density into consideration and present different patterns of two typical attack scenarios, i.e., injecting attack and suspension attack. In addition, we develop a prototype for data acquisition in a controller area network and undertake substantial vehicle experiments. The results show that the attack detection method has advantages in both recognition accuracy and application range compared with the method based on information entropy theory. This research work is expected to contribute to the further development of attacks detection system applied in vehicular networks.