Asynchronous Design of Modular Multiplication Using Adaptive Radix Computation

This paper explores the concurrent property inherent in the Montgomery modular multiplication algorithm and uses the adaptive radix arithmetic to speed up the operations for the proposed local synchronous global asynchronous (LSGA) architecture. The basic idea is to partition the operand into several equal-sized segments and then performs the multiplication and residue calculations in a concurrent LSGA fashion. To further reduce the computation time, multi-bit overlapping scanning technique is employed to decrease the number of operations in each segment under limited hardware resources. Experimental results show that the developed LSGA exhibits much better performance than the synchronous design using the same algorithm