Efficient $k$-Parallel Pipelined NTT Architecture for Post Quantum Cryptography

This paper presents a high efficiency architecture of number theoretic transform (NTT) for ring learning with error (R-LWE) based cryptosystems. A customized multi-path delay feedback (MDF) NTT architecture is proposed with k-parallel pipelined data paths to perform polynomial multiplication effectively. The design was successfully accelerated on Vertex-7 FPGA platform and significantly reduced the cryptosystem's processing time, which achieved approximately 3 times smaller area-time-product (ATP) than previous approaches. The implementation results confirmed the proposed architecture's applicability for high-speed large data security applications.