Higher-Order Time Sharing Masking

Abstract

At CHES 2024, Time Sharing Masking (TSM) was introduced as a novel low-latency masking technique for hardware circuits. TSM offers area and randomness efficiency, as well as glitch-extended PINI security, but it is limited to first-order security. We address this limitation and generalize TSM to higher-order security while maintaining all of TSM’s advantages. Additionally, we propose an area-latency tradeoff. We prove HO-TSM glitch-extended PINI security and successfully evaluate our circuits using formal verification tools. Furthermore, we demonstrate area- and latency-efficient implementations of the AES S-box, which do not exhibit leakage in TVLA on FPGA. Our proposed tradeoff enables a first-order secure implementation of a complete AES-128 encryption core with 92 kGE, 920 random bits per round, and 20 cycles of latency, which does not exhibit leakage in TVLA on FPGA.