### Cross Subkey Side Channel Analysis Based on Small Samples

Fanliang Hu, Huanyu Wang, and Junnian Wang

##### Abstract

The majority of recently demonstrated Deep-Learning Side-Channel Analysis (DLSCA) use neural networks trained on a segment of traces containing operations only related to the target subkey. However, when the size of the training set is limited, as in this paper with only 5K power traces, the deep learning (DL) model cannot effectively learn the internal features of the data due to insufficient training data. In this paper, we propose a cross-subkey training approach that acts as a trace augmentation. We train deep-learning models not only on a segment of traces containing the SBox operation of the target subkey of AES-128 but also on segments for other 15 subkeys. Experimental results show that the accuracy of the subkey combination training model is 28.20% higher than that of the individual subkey training model on traces captured in the microcontroller implementation of the STM32F3 with AES-128. And validation is performed on two additional publicly available datasets. At the same time, the number of traces that need to be captured when the model is trained is greatly reduced, demonstrating the effectiveness and practicality of the method.

Available format(s)
Category
Foundations
Publication info
Published elsewhere. Minor revision.Scientific Reports
DOI
10.1038/s41598-022-10279-9
Keywords
Side-channel attackDeep learningAESCross-subkey traing
Contact author(s)
fanliang @ mail hnust edu cn
History
2022-04-15: revised
See all versions
Short URL
https://ia.cr/2021/1328

CC BY

BibTeX

@misc{cryptoeprint:2021/1328,
author = {Fanliang Hu and Huanyu Wang and Junnian Wang},
title = {Cross Subkey Side Channel Analysis Based on Small Samples},
howpublished = {Cryptology ePrint Archive, Paper 2021/1328},
year = {2021},
doi = {10.1038/s41598-022-10279-9},
note = {\url{https://eprint.iacr.org/2021/1328}},
url = {https://eprint.iacr.org/2021/1328}
}

Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.