RLWE-Oriented High-Speed Polynomial Multiplier Utilizing Multi-Lane Stockham NTT Algorithm
Xiang Peng Feng, Shuguo Li, Sufen Xu · IEEE Transactions on Circuits & Systems II Express Briefs · 2019
Cryptography based on ring learning with error (RLWE) problem has become increasingly popular due to its resistance against quantum analysis. The most time-consuming operation in RLWE cryptosystem is polynomial multiplication. This brief presents a novel polynomial multiplier based on Stockham fast Fourier transform (FFT) algorithm. We propose a multi-lane number theoretic transform (NTT) algorithm which can achieve n-degree polynomial multiplication in (nlgn)/d + 2n/d clock cycles with d lanes of butterfly units. In addition, we also customize a memory addressing strategy and a round constant managing scheme for the proposed multi-lane NTT algorithm. Based on our proposed algorithm, a high-speed polynomial multiplier is accomplished on FPGA platform. Implementation results on Spantan-6 FPGA show that our proposed algorithm can achieve a speed up factor of no less than 2.7 times compared with the state of art designs.