Power Side Channel Resistance of RNS Secure Logic

Over the last decade, significant research effort has gone into secret sharing schemes to secure cryptographic implementations to thwart power side-channel attacks. Higher-order side-channel attacks can correlate the behavior of multiple shares of a bit that leads to learningthe bit state. This violates the power side-channel privacy of cryptographic logic families such as t-privatelogic. The only recourse is to increase the number of secret shares t, which results in excessive hardware (quadratic in t) needs in area, energy and time for providing the desired level of security. In this paper, we present a new secure logic family based on secret sharing concepts using a residue number system. This technique maps the input from binary space into multiple un-correlated shares in the residue domain. These shares are processed independently in independent hardware lanes. The results are decoded back to binary space using the Chinese Remainder theorem. This technique increases the computational complexity for a side channel adversary through proper selection of random mask and residual moduli-which increase both side-channel privacy and cryptographic privacy. Further, we implemented the secure RNS logic and computed the SCA metrics. Finally, we evaluated the power SCA resistance using ML-classifiers. The results show that our RNS secure logic provides better resistance against power side-channel attacks both in terms of power distribution uniformity and success rates of power side channel attack root kits.