Template Attacks, Optimal Distinguishers and the Perceived Information Metric

Sylvain Guilley 1, 2 Annelie Heuser 3, 2, 1 Olivier Rioul 3, 2 François-Xavier Standaert 4
1 SSH - Secure and Safe Hardware
LTCI - Laboratoire Traitement et Communication de l'Information
3 COMNUM - Communications Numériques
LTCI - Laboratoire Traitement et Communication de l'Information
Abstract :

Side-channel analysis is long known as a real threat on unprotected and even protected devices. While template attacks are admittedly the most powerful ones, most practical attacks are of a different kind, such as Kocher’s difference of mean or correlation power analysis. It is the imprecision of the a priori leakage model that accounts for the discrepancy between theory and practice. A closer look into the math- ematical expression of template attacks reveals conditions under which practical distinguishers may approach optimality. A metric aiming at quantifying the amount of leakage in an information-theoretic con- text has been recently put forward as the perceived information. We show that this metric also corresponds to the likelihood computed by a template attack using an imperfect model. We thus consolidate the published literature about optimal (template) attacks and optimal distinguishers by clarifying the state-of-the-art about the strongest side-channel adversarial strategies, with emphasis on the importance of the initial knowledge of the system under attack, which is typically accessed through profiling.

Complete list of metadatas

https://hal.telecom-paristech.fr/hal-02287176
Contributor : Telecomparis Hal <>
Submitted on : Friday, September 13, 2019 - 4:40:34 PM
Last modification on : Thursday, October 17, 2019 - 12:37:03 PM

Identifiers

  • HAL Id : hal-02287176, version 1

Citation

Sylvain Guilley, Annelie Heuser, Olivier Rioul, François-Xavier Standaert. Template Attacks, Optimal Distinguishers and the Perceived Information Metric. 13th International Workshop on Cryptographic Architectures Embedded in Reconfigurable Devices (CryptArchi 2015), Jun 2015, Leuven, Belgium. ⟨hal-02287176⟩

Share

Metrics

Record views

7