Scholarlink

The success of horizontal side-channel attacks depends heavily on the correct extraction of points of interest, which are expected to contain relevant leakages, and on the quality of the traces. If the latter is not sufficient, this will consequently degrade the identification of leakage candidates and often render attacks inapplicable. This work aims to assess the relevance of neural networks in the unsupervised context of horizontal attacks by proposing two methods with alternative objectives to mitigate noise artefacts from the input signal. Their application results in better traces exploitability when using clustering-based horizontal attacks.

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

Improving Leakage Exploitability in Horizontal Side Channel Attacks Through Anomaly Mitigation with Unsupervised Neural Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Gauthier Cler,
  • Sebastien Ordas,
  • Philippe Maurine

摘要

The success of horizontal side-channel attacks depends heavily on the correct extraction of points of interest, which are expected to contain relevant leakages, and on the quality of the traces. If the latter is not sufficient, this will consequently degrade the identification of leakage candidates and often render attacks inapplicable. This work aims to assess the relevance of neural networks in the unsupervised context of horizontal attacks by proposing two methods with alternative objectives to mitigate noise artefacts from the input signal. Their application results in better traces exploitability when using clustering-based horizontal attacks.