Rebuttal to “Sampling Race: Bypassing Timing-based Analog Active Sensor Spoofing Detection on Analog-digital Systems”

Briefly, PyCRA implicitly requires the victim sensor to select a sample rate that is higher than a threshold that depends on the non-zero physical delays present in the attacker. This threshold does not depend on the sampling rate at the attacker, and therefore there is no “race” between the two sample rates. Also, some of the key conclusions in your paper are based on using the simple detector which the PyCRA paper itself had recognized as having weaknesses which were addressed by the more sophisticated Chi-squared detector. Indeed, the PyCRA paper had presented the simple detector not as an alternative solution but as an intermediate step to motivate the development of the Chi-squared detector. Furthermore, your paper portrays PyCRA as requiring a sample interval at the victim sensor that exceeds the physical delays at the attacker, which is contrary to PyCRA’s actual functioning.