Too Easy Fault Injection Attacks on Learning with Rounding (LWR)

Francesco Berti; Sasha Petri; Itamar Levi

Paper 2025/2246

Too Easy Fault Injection Attacks on Learning with Rounding (LWR)

Francesco Berti, Bar-Ilan University

Sasha Petri, TU Darmstadt

Itamar Levi

, Bar-Ilan University

Abstract

We present an extend-and-prune fault Injection attack on serial implementations of Learning With Rounding that drop the least-significant bits. By iteratively isolating and recovering progressively larger key portions via faults, the attack recovers the secret key.

Metadata

Available format(s): PDF
Category: Secret-key cryptography
Publication info: Preprint.
Keywords: cryptography hardware security fault injection learning with error LWR
Contact author(s): francesco berti @ biu ac il
sasha petri @ outlook de
itamar levi @ biu ac il
History: 2025-12-18: approved; 2025-12-13: received; See all versions
Short URL: https://ia.cr/2025/2246
License: CC BY

BibTeX

@misc{cryptoeprint:2025/2246,
      author = {Francesco Berti and Sasha Petri and Itamar Levi},
      title = {Too Easy Fault Injection Attacks on Learning with Rounding ({LWR})},
      howpublished = {Cryptology {ePrint} Archive, Paper 2025/2246},
      year = {2025},
      url = {https://eprint.iacr.org/2025/2246}
}