Paper 2020/1268 Original: eprint.iacr.org/2020/1268 ↗

A Novel Duplication Based Countermeasure To Statistical Ineffective Fault Analysis

Abstract

The Statistical Ineffective Fault Analysis, SIFA, is a recent addition to the family of fault based cryptanalysis techniques. SIFA based attack is shown to be formidable and is able to bypass virtually all the conventional fault attack countermeasures. Reported countermeasures to SIFA incur overheads of the order of at least thrice the unprotected cipher. We propose a novel countermeasure that reduces the overhead (compared to all existing countermeasures) as we rely on a simple duplication based technique. In essence, our countermeasure eliminates the observation that enables the attacker to perform SIFA. The core idea we use here is to choose the encoding for the state bits randomly. In this way, each bit of the state is free from statistical bias, which renders SIFA unusable. Our approach protects against stuck-at faults and also does not rely on any side channel countermeasure. We show the effectiveness of the countermeasure through an open source gate-level fault attack simulation tool. Our approach is probably the simplest and the most cost effective.