A Survey on Sender Identification Methodologies for the Controller Area Network

Abstract

The connectivity of modern vehicles is constantly increasing and consequently also the amount of attack vectors. Researchers have shown that it is possible to access internal vehicle communication via wireless connections, allowing the manipulation of safety-critical functions such as brakes and steering. If a Electronic Control Unit (ECU) can be compromised and is connected to the internal vehicle bus, attacks on the vehicle can be carried out in particular by impersonating other bus participants. Problematic is that the Controller Area Network (CAN), the most commonly used bus technology for internal vehicle communication, does not provide trustworthy information about the sender. Thus ECUs are not able to recognize whether a received message was sent by an authorized sender. Due to the limited applicability of cryptographic measures for the CAN, sender identification methods were presented that can determine the sender of a received message based on physical characteristics. Such approaches can increase the security of internal vehicle networks so that, for example, the manipulations can be limited to a single bus segment, thus preventing the propagation of the attack. This paper presents the different methodologies, which can mainly be divided into the categories time- and voltage-based, identifies problems as well as open questions and compares the existing approaches. The work thus offers an introduction into the topic, identifies possible research fields and enables a quick evaluation of the existing technologies.