You are looking at a specific version 20160924:215731 of this paper. See the latest version.

Paper 2016/922

Leakage Characterizing and Detecting Based on Communication Theory

Wei Yang and Yuchen Cao and Ke Ma and Hailong Zhang and Yongbin Zhou and Baofeng Li

Abstract

Evaluating the side-channel attacks (SCAs) resilience of a crypto device is important and necessary. The SCAs-secure evaluation criteria includes the information theoretic metric and the security metric. The former metric measures the leakage amount of a crypto device. It should be independent with the evaluator. However, the current metrics, e.g. mutual information (MI), conditional entropy and perceived information, are related to the leakage model selected by the evaluator. They only reflect the leakage utilization, rather than the real leakage level of a crypto device. In light of this, we analysis the side-channel as a communication channel and develop two objective metrics, the average MI and capacity of the channel, to characterize the real leakage amount and its upper bound of a crypto device through communication theory. Although the channel capacity is a rough estimation of the leakage amount of the device, it can furnish the leakage amount at the worst case scenario the device may leak. We investigate the estimation methods of the two metric in different noise scenes. Besides, a leakage detection method based on consistency check is developed subsequently. The proposed method are capable of finding the Point-Of-Interests (POIs) in leakage traces and introducing few leakage points cannot be used to mount SCAs. The experiments show the effectiveness of the proposed method.

Metadata
Available format(s)
PDF
Publication info
Preprint. MINOR revision.
Keywords
Side-Channel Analysis
Contact author(s)
generalyzy @ gmail com
History
2019-08-18: withdrawn
2016-09-24: received
See all versions
Short URL
https://ia.cr/2016/922
License
Creative Commons Attribution
CC BY
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.