Side channel analysis and active fault analysis are
now major threats to even mathematically robust cryptographic
algorithms that are otherwise resistant to classical cryptanalysis.
It is necessary to design suitable countermeasures to protect
cryptographic primitives against such attacks. This paper focuses
on designing encryption schemes that are innately secure against
fault analysis. The paper formally proves that one such design
strategy, namely the use of key-dependent SBoxes, is only
partially secure against DFA. The paper then examines the
fault tolerance of encryption schemes that use a key-independent
secret tweak value for randomization. In particular, the paper
focuses on a linear tweak based and a non-linear tweak based
version of a recently proposed block cipher DRECON. The paper
demonstrates that while both versions are secure against classical
DFA, the non-linear tweak based version provides greater fault
coverage against stronger fault models. This fact, together with
the DPA resistance provided by the use of variable S-Boxes,
makes DRECON a strong candidate for the design of secure
cryptographic primitives. All claims have been validated by
experimental results on a SASEBO GII platform.
↧