**Verification of the security in Boolean masked circuits**

*Vahid Jahandideh*

**Abstract: **We introduce a novel method for reducing an arbitrary $\delta$-noisy leakage function to a collection of $\epsilon$-random probing leakages. These reductions combined with linear algebra tools are utilized to study the security of linear Boolean masked circuits in a practical and concrete setting. The secret recovery probability (SRP) that measures an adversary's ability to obtain secrets of a masked circuit is used to quantify the security. Leakage data and the parity-check relations imposed by the algorithm's structure are employed to estimate the SRP

Both the reduction method and the SRP metric were used in the previous works. Here, as our main contribution, the SRP evaluation task is decomposed from the given $\mathbb{F}_q$ field to $q-1$ different binary systems indexed with $i$. Where for the $i$th system, the equivalent $\delta_i$-noisy leakage is reduced optimally to a $\epsilon_i$-random probing leakage with $\epsilon_i=2\delta_i$. Each binary system is targeting a particular bit-composition of the secret. The $q-1$ derived $\delta_i\leq \delta$ values are shown to be a good measure for the informativeness of the given $\delta$-noisy leakage function.

Our works here can be considered as an extension of the work of TCC 2016. There, only $ \delta$-noisy leakage from the shares of a secret was considered. Here, we also incorporate the leakages that are introduced by the computations over the shares.

**Category / Keywords: **implementation / Boolean masking, Noisy leakage, Random probing

**Date: **received 23 Jun 2021

**Contact author: **v jahandideh at gmail com

**Available format(s): **PDF | BibTeX Citation

**Version: **20210624:150526 (All versions of this report)

**Short URL: **ia.cr/2021/860

[ Cryptology ePrint archive ]