系统构建与全面评估：基于Kolmogorov-Smirnov的侧信道鉴别器性能比较
365KB |
更新于2024-07-15
| 20 浏览量 | 举报
收藏
立即下载 本文是一篇深入研究的学术论文，主要关注于通用的旁信道鉴别器在加密模块中的应用，特别是基于Kolmogorov-Smirnov (KS) 测试的方法。Kolmogorov-Smirnov Analysis (KSA) 和 Partial Kolmogorov-Smirnov Analysis (PKS) 是两种在此领域的关键区分器，它们虽然都依赖于KS检验，但在构建策略上有显著差异。
作者们系统地整合了KSA和PKS的构建策略，创造出了九种新的基于KS测试的侧信道鉴别器变体。这一步骤旨在提高区分器的性能和适应性，以便在物理泄漏信息有限的情况下，依然能够有效地揭示嵌入在加密模块中的正确密钥。构建新变体的过程中，他们强调了策略的灵活性和优化。
论文的核心内容是对这十二种鉴别器在不同模拟场景下的综合评估，包括在单变量环境中的有效性与效率研究。通过一系列针对模拟数据和实际数据的攻击，作者们使用成功率 (SR) 作为衡量标准，以确定在具体情境下哪种鉴别器表现最佳。实验结果显示，这些基于KS测试的区分器不仅具有理论上的优越性，还具有在实际应用中的实用价值。
此外，该研究不仅提供了选择合适鉴别器的技术指南，还探讨了这些技术对于保护现代加密系统免受侧信道攻击的重要性。它为密码学社区提供了一个宝贵的工具集，帮助研究人员更好地理解和对抗日益复杂的信息安全威胁。
这篇论文在侧信道分析领域做出了实质性的贡献，展示了如何通过改进和结合现有技术来增强密码安全性，以及在实际环境中评估这些防护措施的有效性。这对于理解和保护现代加密系统免受潜在物理泄露威胁具有重要的实践意义。
Construction and Evaluation of KS Based Distinguishers 339
2.3 PKS Distinguisher
PKS distinguisher is based on single-sample KS test. Its central idea is to measure
the p-value produced by comparing normal distribution and part of conditional
trace distribution L|M . For convenience, leakages L and the hypothetical power
consumptions M are usually processed by Z-score transformation in PKS. p is an
empirical parameter in PKS from zero to one. N(0,1) represents standard normal
distribution. PKS, a two-partial KS test distinguisher, is shown in the formula (6).
D
KS
l
= P
value
(D
KS
(Pr[L = l|M ≤ p]||N (0, 1))) (4)
D
KS
r
= P
value
(D
KS
(Pr[L = l|M>p]||N(0, 1))) (5)
D
PKS
= D
KS
l
× D
KS
r
(6)
PKS will return the smallest product of p-values when the key hypothesis is
correct.
3 Systematic Construction of KS Test Based
Side-Channel Distinguishers
Fromsection2,welearnthatbothKSAandPKSarebasedonKStest,and
they are able to recover the correct key by partitioning the leakages correctly.
However, KSA and PKS are really different from each other in terms of their
construction strategies. Therefore, we will show how to construct other new
variants of KS test based distinguishers by combining their different construction
strategies in a systematic way. For this purpose, we will analyze the construction
strategies using by KSA and PKS, and then we will present nine new variants
of KS test based distinguishers.
3.1 Construction Strategies of KSA and PKS
In this subsection, we will compare the construction differences between KSA
and PKS in four aspects: partition method, similarity measure used by KS test,
assumption about leakages, and normalization.
Partition Method. In a partition attack [16], leakages are divided into sev-
eral sets p
1
k
,p
2
k
, ..., p
n
k
according to each key hypothesis k. These sets are built
according to a power model H. In this paper, partition method is classified as
non-cumulative partition method and cumulative partition method. Examples
of hypothetical leakages that can be used to partition 16-element leakages are
shown in Table 1. Specifically, non-cumulative partition used by KSA is shown
in the left part of Table 1, while cumulative partition used by PKS is shown in
the right part of Table 1.
Similarity Measure Used by KS Test. Distance is used by KSA to measure
the similarity of two distributions. In contrast, p-value is adopted in PKS to
indicate whether or not partial leakages follow a normal distribution.
剩余16页未读，继续阅读
相关推荐
weixin_38724919
- 粉丝: 5
我的内容管理
展开
