Embedded Intrusion Detection System for Detecting Attacks over CAN-BUS

Abstract

The increasing number of connected cars introduced new cyber-attacks strategies that give life to potentially devastating scenarios on everyday life. In fat, the connected cars show many vulnerabilities and are not conform to the policies defined in the AIC Model (availability, integrity and confidentiality). On the other hand, the advantages related to cars connected are very useful for implementing new innovative scenarios providing, for example, context and situation awareness in some operative scenarios. The main problem relies in the introduction of effective techniques that works in well-known framework (PC, Smartphone, …) in a real challenging environment as the automotive. In this scenario, the main parameter to consider is that of the quick ability to identify and react a possible attack. So in this paper, an embedded Intrusion Detection System for Automotive is introduced. It works adopting a Bayesian Network approach for the quick identification of malicious messages on the controller Area Network (CAN-Bus). The first experimental results, obtained in a real scenario, seems to be real interesting.