Towards Private Deep Learning-based Side-Channel Analysis using Homomorphic Encryption

Fabian Schmid; Shibam Mukherjee; Stjepan Picek; Marc Stöttinger; Fabrizio De Santis; Christian Rechberger

Paper 2023/1252

Towards Private Deep Learning-based Side-Channel Analysis using Homomorphic Encryption

Fabian Schmid, Graz University of Technology

Shibam Mukherjee, Graz University of Technology, Know-Center GmbH

Stjepan Picek, Radboud University

Marc Stöttinger, University of Applied Science RheinMain

Fabrizio De Santis

, Siemens AG

Christian Rechberger, Graz University of Technology

Abstract

Side-channel analysis certification is a process designed to certify the resilience of cryptographic hardware and software implementations against side-channel attacks. In certain cases, third-party evaluations by external companies or departments are necessary due to limited budget, time, or even expertise with the penalty of a significant exchange of sensitive information during the evaluation process. In this work, we investigate the potential of Homomorphic Encryption (HE) in performing side-channel analysis on HE-encrypted measurements. With HE applied to side-channel analysis (SCA), a third party can perform SCA on encrypted measurement data and provide the outcome of the analysis without gaining insights about the actual cryptographic implementation under test. To this end, we evaluate its feasibility by analyzing the impact of AI-based side-channel analysis using HE (private SCA) on accuracy and execution time and compare the results with an ordinary AI-based side-channel analysis (plain SCA). Our work suggests that both unprotected and protected cryptographic implementations can be successfully attacked already today with standard server equipment and modern HE protocols/libraries, while the traces are HE-encrypted.

Metadata

Available format(s): PDF
Category: Applications
Publication info: Preprint.
Keywords: Side-channel Analysis Deep Learning Neural Networks Homomorphic Encryption Private AI
Contact author(s): fabian schmid @ iaik tugraz at
shibam mukherjee @ iaik tugraz at
stjepan picek @ ru nl
marc stoettinger @ hs-rm de
fabrizio desantis @ siemens com
christian rechberger @ tugraz at
History: 2023-08-21: revised; 2023-08-18: received; See all versions
Short URL: https://ia.cr/2023/1252
License: CC0

BibTeX

@misc{cryptoeprint:2023/1252,
      author = {Fabian Schmid and Shibam Mukherjee and Stjepan Picek and Marc Stöttinger and Fabrizio De Santis and Christian Rechberger},
      title = {Towards Private Deep Learning-based Side-Channel Analysis using Homomorphic Encryption},
      howpublished = {Cryptology ePrint Archive, Paper 2023/1252},
      year = {2023},
      note = {\url{https://eprint.iacr.org/2023/1252}},
      url = {https://eprint.iacr.org/2023/1252}
}