Paper 2019/305

Practical Algebraic Side-Channel Attacks Against ACORN

Alexandre Adomnicai, Laurent Masson, and Jacques J. A. Fournier

Abstract

The authenticated cipher ACORN is one of the two finalists of the CAESAR competition and is intended for lightweight applications. Because such use cases require protection against physical attacks, several works have been undertaken to achieve secure implementations. Although dedicated threshold and masked schemes have been proposed, no practical side-channel attack against ACORN has been published in the literature yet. It has been theoretically demonstrated that ACORN is vulnerable against differential power analysis but the feasibility of the attack has not been validated in a practical manner. This paper details the results obtained when putting the attack into practice against a software implementation running on a 32-bit micro-controller. Especially, these practical results led us to propose two variants of the reference attack: one that requires less knowledge of initial vectors and another one that is less prone to errors in practice and requires fewer acquisitions.

Note: Fixed a mistake: during the ACORN state update, it is only possible to parallelize up to 33 steps and not 37 as mentioned in the published version.

Metadata
Available format(s)
PDF
Category
Implementation
Publication info
Published elsewhere. Minor revision.ICISC 2018
Keywords
ACORNAEADSide-Channel
Contact author(s)
alex adomnicai @ gmail com
History
2019-03-20: received
Short URL
https://ia.cr/2019/305
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2019/305,
      author = {Alexandre Adomnicai and Laurent Masson and Jacques J. A.  Fournier},
      title = {Practical Algebraic Side-Channel Attacks Against ACORN},
      howpublished = {Cryptology ePrint Archive, Paper 2019/305},
      year = {2019},
      note = {\url{https://eprint.iacr.org/2019/305}},
      url = {https://eprint.iacr.org/2019/305}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.