High-speed polynomial multiplier to accelerate the arithmetical operations of the Post-Quantum Cryptography algorithms

The advent of quantum computers and their increasing computing performance threatens the use of current cryptographic protocols as a way to ensure protection against cyberthreats. For this reason, in 2016, the American National Institute of Standards and Technology started a post-quantum cryptography standardization process for finding new quantum-resistant cryptographic protocols for both key encapsulation mechanisms and digital signatures. Saber is one of the four finalists, it relies on the Module-Learning-with-Rounding problem which is a lattice-based problem and it is believed to be quantum-resistant. The main implementation bottleneck of this protocol is the significant time spent in computing polynomial multiplications in polynomial rings with power of two moduli. This work aims at implementing a hardware architecture that can manage all the arithmetic operations contained in key generation, encryption and decryption functions of the Saber public key encryption protocol, for each of its versions. This is achieved using a schoolbook-based polynomial multiplication accelerator with different optimizations, that rely on centralized multiplication and the smallness of operand polynomials. Results from the design synthesis demonstrate good operating performance and low power dissipation values.