Paper 2023/398

A New Linear Distinguisher for Four-Round AES

Tomer Ashur, Cryptomeria
Erik Takke, Eindhoven University of Technology
Abstract

In SAC’14, Biham and Carmeli presented a novel attack on DES, involving a variation of Partitioning Cryptanalysis. This was further extended in ToSC’18 by Biham and Perle into the Conditional Linear Cryptanalysis in the context of Feistel ciphers. In this work, we formalize this cryptanalytic technique for block ciphers in general and derive several properties. This conditional approximation is then used to approximate the inv : GF(2^8) → GF(2^8) : x → x^254 function which forms the only source of non-linearity in the AES. By extending the approximation to encompass the full AES round function, a linear distinguisher for four-round AES in the known-plaintext model is constructed; the existence of which is often understood to be impossible. We furthermore demonstrate a key-recovery attack capable of extracting 32 bits of information in 4-round AES using 2^125.62 data and time. In addition to suggesting a new approach to advancing the cryptanalysis of the AES, this result moreover demonstrates a caveat in the standard interpretation of the Wide Trail Strategy — the design framework underlying many SPN-based ciphers published in recent years.

Metadata
Available format(s)
PDF
Category
Attacks and cryptanalysis
Publication info
Preprint.
Keywords
Conditional Linear CryptanalysisAESStatistical Distinguisher
Contact author(s)
tomer @ cryptomeria tech
e c takke @ tue nl
History
2023-03-24: approved
2023-03-20: received
See all versions
Short URL
https://ia.cr/2023/398
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2023/398,
      author = {Tomer Ashur and Erik Takke},
      title = {A New Linear Distinguisher for Four-Round {AES}},
      howpublished = {Cryptology {ePrint} Archive, Paper 2023/398},
      year = {2023},
      url = {https://eprint.iacr.org/2023/398}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.