A VLSI architecture of fast high-radix modular multiplication for RSA cryptosystem

This paper presents a high-radix modular multiplication algorithm and its corresponding VLSI architecture for RSA cryptosystem. To reduce the total number of required operations, we partition the multiplier operand into several equal-sized segments and treat each segment as a basic unit for accumulation and module operations. Then, the multiplication and residue calculation of each segment are performed in a pipelined fashion to increase the throughput rate. This paper also shows how to simplify the quotient estimation based on multiple-bit overlapping scanning and to reduce the logic depth in high-radix implementation. Results show that only a small lookup table is needed for quotient estimation in our development and the total operating time is smaller than that of the corresponding radix-2 implementation