Evaluating proposed countermeasures for power analysis of ring oscillator PUFs

The security of PUF derived secrets against common analysis techniques is integral to their use as a hardware security primitive. As in any secure element, it is essential that designers understand the impact that design choices can have on information leakage in various domains. While PUF designs have been well studied in terms of machine learning attacks from a remote adversary, there has been much less investigation of side channel leakage and how to mitigate it. In this paper we investigate different proposed countermeasures to prevent such attacks and the actual degree of protection provided by each. From these results, we show that only one approach (parallel comparison) results in robust protection. This provides a valuable insight into secure design principles for all RO-based PUFs.