An Efficient Hardware Accelerator of High-Speed NTT for CRYSTALS-Kyber Post-Quantum Cryptography

CRYSTALS-Kyber (Kyber) is the sole chosen Public-key Encryption (PKE) and key-establishment algorithm in the finalist round of the Post-quantum Cryptography (PQC) standardization initiated by the National Institute of Standards and Technology (NIST). Within the Kyber protocol, the Number Theoretic Transform (NTT) operation stands out as the most critical and time-intensive component, requiring rapid butterfly units and efficient memory approaches. This paper proposes a high-speed NTT architecture featuring a novel butterfly operation unit tailored for the Kyber protocol, capable of supporting high-frequency operations and effective memory approaches. The proposed NTT structure is realized on the Xilinx Artix-7 FPGA to assess its performance. Experimental results demonstrate that the proposed NTT design can execute the NTT operation in 459 cycles at 314MHz using only 640 LUTs, 667 FFs, and 2 DSPs, offering a more than 24% improvement compared to state-of-the-art designs.