Paper 2022/493

Don’t Learn What You Already Know: Grey-Box Modeling for Profiling Side-Channel Analysis against Masking

Loïc Masure, Valence Cristiani, Maxime Lecomte, and François-Xavier Standaert


Over the past few years, deep-learning-based attacks have emerged as a de facto standard, thanks to their ability to break implementations of cryptographic primitives without pre-processing, even against widely used counter-measures such as hiding and masking. However, the recent works of Bronchain and Standaert at Tches 2020 questioned the soundness of such tools if used in a black-box setting to evaluate implementations protected with higher-order masking. On the opposite, white-box evaluations may be seen as possibly far from what a real-world adversary could do, thereby leading to too conservative security bounds. In this paper, we propose a new threat model that we name grey-box benefiting from a trade-off between black and white box models. Our grey-box model is closer to a real-world adversary, in the sense that it does not need to have access to the random nonces used by masking during the profiling phase like in a white-box model, while it does not need to learn the masking scheme as implicitly done in a black-box model. We show how to combine the power of deep learning with the prior knowledge of grey-box modeling. As a result, we show on simulations and experiments on public datasets how it allows to reduce by an order of magnitude the profiling complexity, i.e., the number of profiling traces needed to satisfyingly train a model, compared to a fully black-box model.

Note: Adding Acknowledgements

Available format(s)
Publication info
Deep LearningSide Channel AnalysisMaskingBlack BoxGrey BoxProfiling
Contact author(s)
loic masure @ uclouvain be
2022-04-23: received
Short URL
Creative Commons Attribution


      author = {Loïc Masure and Valence Cristiani and Maxime Lecomte and François-Xavier Standaert},
      title = {Don’t Learn What You Already Know: Grey-Box Modeling for Profiling Side-Channel Analysis against Masking},
      howpublished = {Cryptology ePrint Archive, Paper 2022/493},
      year = {2022},
      note = {\url{}},
      url = {}
Note: In order to protect the privacy of readers, does not use cookies or embedded third party content.