Is information leakage spilling? - Université Jean-Monnet-Saint-Étienne Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Is information leakage spilling?

Lilian Bossuet
Vincent Grosso
Carlos Andres Lara-Nino

Résumé

Correlation Power Analysis (CPA) is one of the best known approaches for performing power analysis on cryptographic algorithms. If the implementation is unprotected, a reasonable number of power traces is required to validate the optimal key hypotheses and probably retrieve some part of a secret key. This is achieved by computing a correlation matrix with as many columns as the total number of key guesses and as many rows as the number of samples in the power traces. Then, the correct hypothesis is chosen as the guess with the largest correlation coefficient (or the largest absolute, depending on who you ask). But this ignores the second dimension of the matrix since we assume that a single sample will exhibit the greatest correlation against the model. However, what happens when the correlation “spills” over multiple samples? In this talk we show that in such a case it is possible to use an area estimator to improve the accuracy of selecting the correct hypotheses. We also discuss our impressions on how is this spill can be achieved.
Fichier non déposé

Dates et versions

ujm-04220174 , version 1 (27-09-2023)

Identifiants

  • HAL Id : ujm-04220174 , version 1

Citer

Lilian Bossuet, Vincent Grosso, Carlos Andres Lara-Nino. Is information leakage spilling?. Cryptographic architectures embedded in logic devices, Jun 2023, Castro Urdiales, Spain. ⟨ujm-04220174⟩
17 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More