Maskara: Compilation of a Masking Countermeasure with Optimised Polynomial Interpolation - Sorbonne Université Access content directly
Journal Articles IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems Year : 2020

Maskara: Compilation of a Masking Countermeasure with Optimised Polynomial Interpolation

Abstract

Side-channel attacks are amongst the major threats for embedded systems and IoT devices. Masking is one of the most used countermeasure against such attacks, but its application remains a difficult process. We propose a target-independent approach for applying a first-order boolean masking countermeasure during compilation, on the static single assignment form. Contrary to state-of-the art automated approaches that require to simplify the control flow of the input program, our approach supports regular control-flow program structures. Moreover, our compiler is the first to automatically mask table lookups using a polynomial interpolation approach. We also present new optimisations to speed up the evaluation of polynomials: we reduce the number of terms of the polynomial, and we accelerate finite field multiplication. We show that our approach is faster than the standard masked table approach with mask refresh after each access, with speedups up to ×2.4 in our experiments. Finally, using a formal verification approach, we show that the compiled machine code is secure, i.e., that all intermediate computations are statistically independent of the secrets.
Fichier principal
Vignette du fichier
Maskara.pdf (446.26 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-02931632 , version 1 (02-03-2021)

Identifiers

Cite

Nicolas Belleville, Damien Couroussé, Karine Heydemann, Quentin Meunier, Inès Ben El Ouahma. Maskara: Compilation of a Masking Countermeasure with Optimised Polynomial Interpolation. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2020, 39 (11), pp.1-1. ⟨10.1109/TCAD.2020.3012237⟩. ⟨hal-02931632⟩
216 View
231 Download

Altmetric

Share

Gmail Facebook X LinkedIn More