Novel methodology to improve Multi-moduli architectures for Binary-to-RNS conversion

Abstract

This paper presents a novel approach to improve the existing Binary-to-RNS multi-moduli architectures (MMAs). These MMAs reduce the complexity by sharing common intermediate circuitry among various RNS moduli channels. Two types of MMAs are distinguished depending on whether the functionality is implemented in serial or parallel. An existing input pre-computation methodology, which improves the performance of the final MMA modular conversion of our approach herein first time used. Experimental results suggest that the proposed RNS design accelerates the execution, compensating the additional cost incurred in power and area. Moreover, the designs herein proposed have demonstrated energy reductions.