Hardware implementation of Ring-LWE lattice cryptography with BCH and Gray coding based error correction

Lattice structure based cryptographic algorithms are well-known in Post-Quantum-Cryptography (PQC) due to their strong security against quantum computers, better performance and practicality in hardware implementation. Among popular Ring learning with errors (RLWE) lattice problem based cryptosystems, NewHope Simple is a promising key encapsulation mechanism (KEM). The major challenges of the algorithm are timeXarea complexity, bandwidth utilization, and a trade-off between security level and decryption failure rate (DFR). In this paper, we propose a timeXspace optimized hardware implementation of the NewHope Simple KEM along with Bose-Chaudhuri-Hocquenghem (BCH) and weighted Gray coding based error correction technique. The design is validated on Zynq7000 Artix-7 FPGA. The proposed NewHope Simple design achieves 5.03X improvement in areaXtime product than other hardware implementations, and at least 12% enhancement in execution time compared to various Graphics Processing Unit (GPU) based implementations. The result shows that our BCH decoder requires 5.8X lesser computation time than its counterparts. Our error correction technique achieves 10.4% better error rate than BCH for 4 bits per symbol. This improvement in bit error rate (BER) can be exploited to enhance bandwidth utilization and security. The proposed architecture can complete a successful key exchange in 634μs.