• 융합보안논문지
• /
• 제2권2호
• /
• pp.143-153
• /
• 2002

In this paper, we propose a multiplication scheme based on cellular automata and propose high speed multiplication scheme and exponentiation scheme using a optimal normal basis. And then EIGamal signature scheme is implemented by proposed schemes. A proposed multiplication and exponentiation scheme based on cellular automata can be used in restricted computing environments such that basis is frequently changed and cryptosystem and multimedia applications that are required high speed operations.

• 정보보호학회논문지
• /
• 제31권3호
• /
• pp.473-480
• /
• 2021

차세대 공개키 암호 고속 연산을 위해서는 목표로 하는 컴퓨터 프로세서의 구조를 활용하여 암호화 기본 연산을 최적화 구현하는 것이 중요하다. 본 논문에서는 RISC-V 프로세서 상에서 차세대 공개키 암호 고속 연산을 위해 핵심 곱셈기 연산을 최적화 구현하는 기법을 제안한다. 특히 RISC-V 프로세서의 기본 연산자를 열 기반 곱셈기 연산알고리즘에 맞추어 최적 구현해봄으로서 이전 연구와 비교 시 256-비트 곱셈의 경우 약 19% 그리고 512-비트 곱셈의 경우 약 8%의 성능 향상을 RISC-V 프로세서 상에서 달성하였다. 마지막으로 RISC-V 프로세서에서 추가적으로 제공되면 곱셈 연산 성능 향상에 도움이 될 수 있는 확장 명령어 셋에 대해서도 확인해 보도록 한다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 하계종합학술대회 논문집(2)
• /
• pp.173-176
• /
• 2002

This paper is about a high-speed MAC (multiplier and accumulator) design applying radix-4 and radix-8 Booth's algorithm at the same time. The optimized hybrid radix design for high speed MAC has taken advantage of both a radix-4 and a radix-8 architectures. A radix-4 architecture meets high-speed, but it takes much more power and chip area than a radix-8 architecture. A radix-8 architecture needs less power and chip area than the other, but it has a bottleneck of generating three times the multiplicand problem. An optimized hybrid architecture performs the radix-4 multiplication partially in parallel with the generation of three times the multiplicand for use of the radix-8 multiplication. It reduces the concerned bit width of multiplier in radix-8 multiplication.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 하계종합학술대회 논문집(1)
• /
• pp.125-128
• /
• 2002

This paper is about a high-speed MAC (multiplier and accumulator) design applying radix-4 and radix-8 Booth's algorithm at the same time. The optimized hybrid radix design for high speed MAC has taken advantage of both a radix-4 and a radix-8 architectures. A radix-4 architecture meets high-speed, but it takes much more power and chip area than a radix-8 architecture. A radix-8 architecture needs less power and chip area than the other, but it has a bottleneck of generating three times the multiplicand problem. An optimized hybrid architecture performs tile radix-4 multiplication partially in parallel with the generation of three times the multiplicand for use of tile radix-8 multiplication. It reduces the concerned bit width of multiplier in radix-8 multiplication.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2005년도 춘계종합학술대회
• /
• pp.925-927
• /
• 2005

The study is about a finite field multiplying unit, which performs a calculation t-times as fast as the Mastrovito's multiplier architecture, suggesting and using the 2-times faster multiplier architecture. Former studies on finite field multiplication architecture includes the serial multiplication architecture, the array multiplication architecture, and the hybrid finite field multiplication architecture. Mastrovito's serial multiplication architecture has been regarded as the basic architecture for the finite field multiplication, and in order to exploit parallelism, as much resources were expensed to get as much speed in the finite field array multipliers. The array multiplication architecture has weakness in terms of area/performance ratio. In 1999, Parr has proposed the hybrid multipcliation architecture adopting benefits from both architectures. In the hybrid multiplication architecture, the main hardware frame is based on the Mastrovito's serial multiplication architecture with smaller 2-dimensional array multipliers as processing elements, so that its calculation speed is fairly fast costing intermediate resources. However, as the order of the finite field, complex integers instead of prime integers should be used, which means it cannot be used in the high-security applications. In this paper, we propose a different approach to devise a finite field multiplication architecture using Mastrovito's concepts.

• Journal of information and communication convergence engineering
• /
• 제15권4호
• /
• pp.212-216
• /
• 2017

In this paper, we introduce novel techniques to improve the high performance of AE functions on modern high-end IoT platforms (ARM-NEON), which support SIMD and cryptography instruction sets. For the Sophie Germain Counter Mode of operation (SGCM), counter modes of encryption and prime field multiplication are required. We chose the Montgomery multiplication for modular multiplication. We perform Montgomery multiplication in a parallel way by exploiting both the ARM and NEON instruction sets. Specifically, the NEON instruction performed 128-bit integer multiplication and the ARM instruction performed Montgomery reduction, simultaneously. This approach hides the latency for ARM in the NEON instruction set. For a high-speed counter mode of encryptions for both AE functions, we introduced two-level computations. When the tasks were large volume, we switched to the NEON instruction to execute the encryption operations. Otherwise, we performed the encryptions on the ARM module.

• ETRI Journal
• /
• 제19권4호
• /
• pp.317-325
• /
• 1997

A new on-the-fly conversion algorithm is proposed, and high-speed array multipliers with the on-the-fly conversion are presented. The new on-the-fly conversion logic is used to speed up carry-propagate addition at the last stage of multiplication, and provides constant delay independent of the number of input bits. In this paper, the multiplication architecture and the on-the-fly conversion algorithm are presented and discussed in detail. The proposed architecture has multiplication time of (n +1)$t_{FA}$, Where n is the number of input bits and $t_{FA}$ is the delay of a full adder. According to our comparative performance evaluation, the proposed architecture has shorter delay and requires less area than the conventional array multiplier with on-the-fly conversion.

• 한국정보통신학회논문지
• /
• 제14권3호
• /
• pp.628-636
• /
• 2010

본 논문에서는 GF($2^m$)상의 표준기저를 사용한 새로운 형태의 VCG에 의한 고속병렬 승산회로를 제안하였다. 승산기의 구성에 앞서, 피승수 다항식과 기약다항식의 승산을 병렬로 수행하는 벡터 코드 생성기(VCG) 기본 셀을 설계하였고, VCG 회로와 승수 다항식의 한 계수와 비트-병렬로 승산하여 결과를 생성하는 부분 승산결과 셀(PPC)를 설계하였다. 제안한 승산기는 VCG와 PPC를 연결하여 고속의 병렬 승산을 수행한다. VCG 기본 셀과 PPC는 각각 1개의 AND 게이트와 1개의 XOR 게이트로 구성된다. 이러한 과정을 확장하여 m에 대한 일반화된 회로의 설계를 보였으며, 간단한 형태의 승산회로 구성의 예를 GF($2^4$)를 통해 보였다. 또한 제시한 승산기는 PSpice 시뮬레이션을 통하여 동작특성을 보였다. 본 논문에서 제안한 승산기는 VCG와 PPC을 반복적으로 연결하여 구성하므로, 차수 m이 매우 큰 유한체상의 두 다항식의 곱셈에서 확장이 용이하며, VLSI에 적합하다.

• Journal of the Optical Society of Korea
• /
• 제18권5호
• /
• pp.611-615
• /
• 2014

We report high-speed pulse train generation from a relatively low-speed 10-GHz mode-locked laser by means of line-by-line spectral coding. To increase the pulse repetition rate multiplication (RRM) factor, we combine coding schemes for both spectral intensity and phase by placing a simple mask at the coder focal plane. The resulting RRM factor, determined by multiplying the RRM factors of the individual coding schemes, rises as high as 16. To verify the generated pulses, the optical spectra and autocorrelation traces are examined.

• 한국통신학회논문지
• /
• 제27권3C호
• /
• pp.256-262
• /
• 2002

본 논문에서는 sign estimation technique (3)을 기초로 한 radix-4 모듈라 곱셈 알고리즘을 제안한다. Sign estimation technique은 carry와 sum의 형태로 표현되는 수에서 부호를 알아내는 것이다. 이 방법은 5비트 carry look-ahead adder로 구현이 가능하다. RSA와 같은 암호화 시스템에서는 모듈라 곱셈이 하드웨어의 성능을 좌우한다. 제안한 알고리즘은 modulus가 n 비트인 경우, 모듈라 곱셈 수행시 일반적인 알고리즘의 약 반 클럭 (n/2+3) 사이클만 필요하다. 그래서 매우 큰수의 modulus 사용하는 RSA 암호시스템에서 모듈라 멱승 연산에 매우 효율적이다. 또한 모듈라 곱셈의 하드웨어 성능을 향상하기 위해, CSA (Carry Save Adder)의 맨 마지막 출력에 사용되는 CPA (Carry Propagation Adder) 대신 고속 덧셈기(7)를 사용하였다. 모듈라 멱승 계산이 n 클럭이 소요되는 RL binary 방법을 적용하여 1024 비트 데이터를 RSA 암호화하는데 n(n/2+3) 클럭 사이클만 소요된다.