A flexible RNS-based large polynomial multiplier for Fully Homomorphic Encryption

In the era of the cloud computing, homomorphic encryption allows remote data processing while preserving confidentiality. Its main drawback, however, is the huge complexity in terms of operand size and computation time, which makes hardware acceleration desirable in order to achieve acceptable performance. In this paper, we present a flexible modular polynomial multiplier implemented through a high-level synthesis flow. We show that flexibility does not come at a price, and the proposed solution is competitive against custom designs.