From a Bit-flipping to a Vulnerability in the CAN Standard
Published on: August 16, 2017
TL;DR: CAN-based protocols are vulnerable to bit-flipping attacks at the link layer. In this collaborative research, Politecnico di Milano’s NECSTLab and Trend Micro’s FTR analyze the protocol in depth and demonstrate the vulnerability on a real car, with PoC and so on.
This project started somewhere between 2015 and 2016. Back then, I was an Assistant Professor at Politecnico di Milano. Together with my colleague Stefano, I was advising this bright Master student, Andrea Palanca, who was basically “breathing cars,” and really passionate about car hacking. So it made a lot of sense to introduce him to Eric Evencick, who then became part of the project.
Eric had this idea of looking at the link layer of the CAN protocol, given the numerous frame-injection attacks popping up here and there. However, we’ve thought, injecting frame in the CAN bus is pretty noisy, and also quite easy to detect with some smart monitoring. Indeed, knowing the network architecture and ECUs attached to it (and, let’s be honest, they don’t change that often even in a connected car), it’s fairly easy to figure out if a frame is out of order, or simply unexpected.
Andrea then started to dig into the CAN bus standard, using his own car as a playground, and quickly came up with a prototype testbed CAN deployment in the lab, on which he started to explore the effect of flipping “the right bit at the right time”.
Alright, enough high-level talking! It’s time to take a look at the actual content.
Gimme da Video!
In the following short video you can see me presenting the work to an academic audience. True fact: I’ve recorded this while preparing my talk for DIMVA 2017.
If you’re curious to know more:
for an intermediate-level description, you can read our blog post,
A Stealth, Selective, Link-Layer Denial-of-Service Attack Against Automotive Networks
Andrea Palanca, Eric Evenchick, Federico Maggi, and Stefano Zanero.
In Michalis Polychronakis & Michael Meier, eds. Proceedings of the 14th International Conference on Detection of Intrusions and Malware, and Vulnerability Assessment (DIMVA). Bonn, Germany: Springer International Publishing, 185–206. DOI: http://dx.doi.org/10.1007/978-3-319-60876-1_9 (July 6, 2017) [PDF]