Fast, compact and secure implementation of rsa on dedicated hardware

Official URL: http://192.168.1.20/record=b1228179 (Table of Contents)

RSA is the most popular Public Key Cryptosystem (PKC) and is heavily used today. PKC comes into play, when two parties, who have previously never met, want to create a secure channel between them. The core operation in RSA is modular multiplication, which requires lots of computational power especially when the operands are longer than 1024-bits. Although today’s powerful PC’s can easily handle one RSA operation in a fraction of a second, small devices such as PDA’s, cell phones, smart cards, etc. have limited computational power, thus there is a need for dedicated hardware which is specially designed to meet the demand of this heavy calculation. Additionally, web servers, which thousands of users can access at the same time, need to perform many PKC operations in a very short time and this can create a performance bottleneck. Special algorithms implemented on dedicated hardware can take advantage of true massive parallelism and high utilization of the data path resulting in high efficiency in terms of both power and execution time while keeping the chip cost low. We will use the “Montgomery Modular Multiplication” algorithm in our implementation, which is considered one of the most efficient multiplication schemes, and has many applications in PKC. In the first part of the thesis, our “2048-bit Radix-4 based Modular Multiplier” design is introduced and compared with the conventional radix-2 modular multipliers of previous works. Our implementation for 2048-bit modular multiplication features up to 82% shorter execution time with 33% increase in the area over the conventional radix-2 designs and can achieve 132 MHz on a Xilinx xc2v6000 FPGA. The proposed multiplier has one of the fastest execution times in terms of latency and performs better than (37% better) our reference radix-2 design in terms of time-area product. The results are similar in the ASIC case where we implement our design for UMC 0.18 μm technology. In the second part, a fast, efficient, and parameterized modular multiplier and a secure exponentiation circuit intended for inexpensive FPGAs are presented. The design utilizes hardwired block multipliers as the main functional unit and Block-RAM as storage unit for the operands. The adopted design methodology allows adjusting the number of multipliers, the radix used in the multipliers, and number of words to meet the system requirements such as available resources, precision and timing constraints. The deployed method is based on the Montgomery modular multiplication algorithm and the architecture utilizes a pipelining technique that allows concurrent operation of hardwired multipliers. Our design completes 1020-bit and 2040-bit modular multiplications* in 7.62 μs and 27.0 μs respectively with approximately the same device usage on Xilinx Spartan-3E 500. The multiplier uses a moderate amount of system resources while achieving the best area-time product in literature. 2040-bit modular exponentiation engine easily fits into Xilinx Spartan-3E 500; moreover the exponentiation circuit withstands known side channel attacks with an insignificant overhead in area and execution time. The upper limit on the operand precision is dictated only by the available Block-RAM to accommodate the operands within the FPGA. This design is also compared to the first one, considering the relative advantages and disadvantages of each circuit.