A practical differential fault attack on ML-DSA and HAETAE, showing how public coefficients can affect key recovery in post-quantum signature implementations.
This paper presents a practical differential fault attack against ML-DSA and HAETAE.
It focuses on the role of public coefficients in lattice-based signature schemes and shows how faulty signatures can be used for key recovery.
Key Points
Studies practical differential fault attacks on post-quantum digital signatures.
Targets ML-DSA and HAETAE implementations.
Highlights the role of public coefficients in fault-based key recovery.
@misc{shin2026publiccoefficient,
title={Public Coefficient Matters: A Practical Differential Fault Attack on ML-DSA and HAETAE},
author={Shin, WonGeun and Jeon, SeungHyeon and Bae, Daehyeon and Park, Sujin and Kim, HeeSeok},
howpublished={IACR Cryptology ePrint Archive, Report 2026/1344},
year={2026},
url={https://eprint.iacr.org/2026/1344}
}
This paper studies first-order masking for neural-network components that are difficult to protect against side-channel leakage, especially activation and ArgMax gadgets.
The work targets side-channel resistant neural network implementations and provides practical masked building blocks for secure inference.
Key Points
Designs first-order masked activation and ArgMax gadgets.
Focuses on side-channel resistant neural network implementations.
Provides implementation-oriented countermeasures for cryptographic and embedded-security settings.
Publication Info
Journal: IACR Transactions on Cryptographic Hardware and Embedded Systems
Volume / Issue / Pages: 2026 / 2 / 106–131
Published:
Authors: Won Geun Shin, JeongHwan Lee, Sangyun Jung, HeeSeok Kim
DOI: 10.46586/tches.v2026.i2.106-131
@article{shin2026armor,
title={{ARMOR}: First-Order Masking of Activation and {ArgMax} Gadgets for Side-Channel Resistant Neural Networks},
author={Shin, Won Geun and Lee, JeongHwan and Jung, Sangyun and Kim, HeeSeok},
journal={IACR Transactions on Cryptographic Hardware and Embedded Systems},
volume={2026},
number={2},
pages={106--131},
year={2026},
month=apr,
doi={10.46586/tches.v2026.i2.106-131},
url={https://tches.iacr.org/index.php/TCHES/article/view/12882}
}
Conference paper. Bibliographic metadata is registered by IEEE under DOI 10.1109/SVCC61185.2024.10637306.
@inproceedings{chae2024password,
title={The Threat of Password Guessing Attacks Exploiting Linguistic Characteristics: A Case Study on the {Korean} Domains},
author={Chae, MinAh and Shin, WonGeun and Jung, SangYun and Yeom, JiHun and Jeon, DongHo and Kim, HeeSeok},
booktitle={2024 Silicon Valley Cybersecurity Conference (SVCC)},
pages={1--4},
publisher={IEEE},
year={2024},
doi={10.1109/SVCC61185.2024.10637306},
url={https://doi.org/10.1109/SVCC61185.2024.10637306}
}
For Korean-language publications, English titles follow the CV; Korean titles follow the linked publication records or conference programs.
Listed as paper 265 in the official conference program. The linked document is the program, not the paper full text.
@inproceedings{shin2025leakage,
title={누적 분산 알고리즘을 통한 부채널 누수 평가 최적화},
author={신원근 and 전승현 and 김희석},
booktitle={2025년 한국정보보호학회 동계학술대회 (CISC-W 2025)},
year={2025},
note={In Korean; English title in author CV: Optimizing Side-Channel Leakage Assessment with Accumulated Distributed Algorithms},
url={https://di0zxmb8pwajl.cloudfront.net/kiisc/conference/ciscs2025/ciscw2025_programbook_v6.pdf}
}
Korean-language journal article. Title, author list, volume, issue, and page range follow the KoreaScience record.
@article{lee2023lwe,
title={{LWE}-like {KEM}/전자서명에 대한 부채널 대응기법 동향},
author={Lee, JeongHwan and Jung, Sang-Yun and Shin, Won Geun and Park, Sujin and Kim, HeeSeok},
journal={Review of KIISC},
volume={33},
number={3},
pages={85--96},
year={2023},
note={In Korean; English title in author CV: Trends in Side-Channel Countermeasures for LWE-like KEM and Digital Signature Schemes},
url={https://koreascience.kr/article/JAKO202321580231031.page}
}
Listed in the AI Security 2 oral session of the official conference program. The linked document is the program, not the paper full text.
@inproceedings{shin2023binary,
title={이진 신경망 보호를 위한 마스킹 기반 부채널 분석 대응기술},
author={신원근 and 정상윤 and 김희석},
booktitle={2023년 한국정보보호학회 하계학술대회 (CISC-S 2023)},
year={2023},
note={In Korean; English title in author CV: Side-Channel Countermeasure for Binary Neural Network Protection},
url={https://kiisc.or.kr/bbs/downloadBoardImage?uploadedFileId=8053}
}