Scholarlink

This paper target’s the ANSSI’s protected AES implementation on STM32 platform with ARM cortex-M architecture. The dataset generated using this implementation is available publicly as ASCADv2 [11]. By using this available dataset, we targeted ANSSI’s protected AES implementation using multi-task learning (MTL) approach, introduced by Magrehbi in 2020 [17]. Our deep learning based approach is different in terms of model selection, selection of point of interest and targeted operation for attacking as compared to the released paper by Loïc Masure and Rémi Strullu in year 2023 [12]. We proposed two separate MTL Multilayer Perceptron (MLP) neural networks that are used for recovering the complete key byte of the given implementation in 40 traces.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

MLP is Better than ResNet on ANSSI’s Protected AES Implementation on ARM

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Akash Gupta,
  • S. P. Mishra,
  • Atul Prakash

摘要

This paper target’s the ANSSI’s protected AES implementation on STM32 platform with ARM cortex-M architecture. The dataset generated using this implementation is available publicly as ASCADv2 [11]. By using this available dataset, we targeted ANSSI’s protected AES implementation using multi-task learning (MTL) approach, introduced by Magrehbi in 2020 [17]. Our deep learning based approach is different in terms of model selection, selection of point of interest and targeted operation for attacking as compared to the released paper by Loïc Masure and Rémi Strullu in year 2023 [12]. We proposed two separate MTL Multilayer Perceptron (MLP) neural networks that are used for recovering the complete key byte of the given implementation in 40 traces.