Constructive Side-Channel Analysis and Secure Design

Masking schemes achieve provable security against sidechannel analysis by using secret sharing to decorrelate key-dependent intermediate values of the cryptographic algorithm and side-channel information. Masking schemes make assumptions on how the underlying leakage mechanisms of hardware or software behave to account for various physical effects. In this paper, we investigate the effect of the physical placement on the security using leakage assessment on power measurements collected from an FPGA. In order to differentiate other masking failures, we use threshold implementations as masking scheme in conjunction with a high-entropy pseudorandom number generator. We show that we can observe differences in—possibly—exploitable leakage by placing functions corresponding to different shares of a cryptographic implementation in close proximity.