Yet Another Physical Leakage Assessment with the Wasserstein Distance

Abstract

Side-channel attacks pose a significant threat to the security of cryptographic systems by exploiting information leakage such as power consumption. Evaluating the difficulty of such attacks is crucial for developing robust countermeasures. In this study, we propose an analytical methodology using the Wasserstein distance as a metric to assess the vulnerability of cryptographic implementations for power side-channel attacks. We conduct both simulated and FPGA-based experiments to validate the effectiveness of using the Wasserstein distance. Our results demonstrate a clear correlation between the signal-to-noise ratio and the Wasserstein distance, with higher noise leading to smaller distances, thus indicating increased attack difficulty. Notably, the metric stabilizes with a relatively small number of traces, suggesting practical applicability. These findings suggest that the Wasserstein distance can serve as a reliable indicator of attack difficulty, providing a quantifiable measure to aid in the development and evaluation of secure cryptographic systems.