Impact of Wiring Characteristics on Voltage-based Fingerprinting in Controller Area Networks

Abstract

Voltage patterns generated by Controller Area Network (CAN) nodes have been commonly proposed as a source for sender identification as this exposes fine grain manufacturing characteristics. However, the influence of wiring on voltage patterns was insufficiently studied so far and it may be critical in understanding the accuracy of the fingerprinting process. Here we study the influence of wiring on three voltage characteristics: slew rate distribution of recessive to dominant transitions, peak-to-peak and peak-to-root mean square distributions on the plateau area of a dominant bit. Using collected voltage data, we identify slew rate variations depending on the wiring used in each experimental setup. Voltage patterns collected in a laboratory setup with automotive grade cables seem to be identical with those from real-world vehicles, which suggests that this type of cables should be used for realistic experiments.