Data and Computational Fault Detection Mechanism for Devices That Perform Modular Exponentiation

摘要

Fault attacks have become an efficient methodology for extracting secrets stored in embedded devices, and proper countermeasures against such attacks are nowadays considered necessary. This paper describes a simple method for foiling transient fault attacks on devices that perform modular exponentiation with a secret exponent. In the considered scenario, acknowledging an error only at the end of the computations leaks out secret information, and should be avoided. To tackle this difficulty, we propose a scheme that checks, independently, each step (i.e., multiplication/squaring) of the exponentiation algorithm, and aborts the procedure as soon as an error is detected, without completing the computation.