Efficient Profiled Attacks on Masking Schemes

Abstract

Side-channel adversaries represent real-world threats against (certified and uncertified) cryptographic devices. Masking schemes represent prevailing countermeasures to reduce the success probabilities of side-channel attacks. However, masking schemes increase the implementation cost in term of power consumption, clock cycles, and random number generation. Investigation of tools evaluating the degree of resilience of cryptographic devices using masking (against side-channel attacks) represents an important aspect in certification procedures (e.g., Common Criteria, FIPS 140-2 and EMVco). Several side-channel evaluation techniques exist such as template attacks and machine learning based attacks. In this paper, we formalise results obtained in side-channel attacks community when targeting masked implementations. We report theoretical as well as practical results of parametric and non-parametric side-channel attacks on masking schemes. The theoretical part reports results based on a simulation of the execution of software devices while the practical part focuses on actual leakages measured during the execution of a software implementation in three different contexts (that contain different levels of noise).

Efficient Profiled Side-Channel Analysis of Masked Implementations, Extended

“…• Easier to interpret metrics. Lerman et al's analyzes compare profiling methods by first fixing a number of profiling and attack measurements followed by an evaluation of the attacks' probability of success [45].…”

“…We further contextualize the attack settings based on adversarial capabilities and consider profiled attacks with known or unknown masking randomness. • Additional distinguishers The authors of [45] evaluate Gaussian TAs, Kernel Density Estimation (KDE) and Random Forests (RFs), but some natural candidates for the efficient profiling of masked implementations are still missing, for example the Expectation-Maximization (EM) algorithm to profile Gaussian mixtures [28], [50], Soft Analytical Side-Channel Attacks (SASCA) [51] and Multi-Layer Perceptrons (MLP) [52], [53]. We include these distinguishers in our experiments.…”

“…Our experimental analyzes are based on simulated leakages. The rationale behind this choice is similar to the one of Lerman and Markowitch [45]. Since we aim to compare distinguishers, it allows us to evaluate them in different well-controlled scenarii (e.g., low-noise, high-noise), which considerably simplifies the interpretation of the results.…”

“…Before performing quantitative (experimental) comparisons in the next section, we propose a systematic contextualization of the distinguishers we study. We consider both the tools studied by Lerman and Markowitch in [45] and the new ones we list in subsection 2.4. The resulting classification is sumarized in Table 1.…”

“…Whenever trying to estimate the worstcase security level, the profiled setting is preferable, since there is no generic attack strategy that can succeed against any device without profiling [44]. This has led Lerman and Markowitch to initiate a study of efficient profiled attacks against masking schemes as a central ingredient for the security evaluation of cryptographic implementations, and to propose a systematic comparison of parametric and nonparametric distinguishers for this purpose [45].…”