High-Performance Multi-RNS-Assisted Concurrent RSA Cryptosystem Architectures

Abstract

In public-key cryptography, the RSA algorithm is an inevitable part of hardware security because of the ease of implementation and security. RSA Cryptographic algorithm uses many modular arithmetic operations that decide the overall performance of the architecture. This paper proposes VLSI architecture to implement an RSA public-key cryptosystem driven by the Residue Number System (RNS). Modular exponentiation in the RSA algorithm is executed by dividing the entire process into modular squaring and multiplication operations. Based on the RNS employment in modulo-exponential operation, two RSA architectures are proposed. A Verilog HDL code is used to model the entire RSA architecture and ported in Zynq FPGA (XC7Z020CLG484-1) for Proof of Concept (PoC). The Cadence Genus Synthesizer tool characterizes a system’s performance for TSMCs standard Cell library. Partial RNS (Proposed-I)- and Fully RNS (Proposed-II)-based RSA architectures increase the operation speed by 13% and 35%, respectively, compared with the existing RSA. Even though there is an increase in parameters like area, power and PDP for a smaller key size, the improvement in area utilization and encryption/ decryption speed of RSA for a larger key size is evident from the analysis.