Safe-Error Attack
Safe-error attacks recover the key bit by bit by observing whether an injected fault affects the output (a "safe" fault leaves it unchanged), without differential analysis.
Purpose
They offer a simple, efficient fault attack on public-key implementations such as RSA, needing only a changed/unchanged output observation.
Mechanism
A fault is injected into an intermediate value of the RSA-CRT computation; if the final output is unchanged, the fault hit an "inactive" branch (key bit 0); if it changes, the branch was "active" (key bit 1), recovering the key bit by bit.
Use cases
- RSA-CRT fault attacks
- Public-key implementation security assessment
- Smart-card attacks