• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2005년도 추계종합학술대회
• /
• pp.816-819
• /
• 2005

자바 프로그램의 실행 환경인 자바가상기계에서는 거의 모든 바이트코드들이 메모리 상에서의 자료 이동을 필요로 한다. 자료 이동은 에너지 소비를 일으키므로 이것의 최소화는 JVM의 에너지 효율성 향상에 매우 중요하다. 특히 자바 반복문은 반복 회수에 비례하는 양만큼의 자료 이동을 요구하므로 JVM의 에너지 효율을 높이기 위해서는 무엇보다도 자바 반복문의 에너지 효율을 높이는 것이 중요하다. 본 논문에서는 자바 반복문을 바이트코드 수준에서 분석하여 에너지 효율성을 높일 수 있는 여러 방법들을 제안하였다.

• 정보처리학회논문지C
• /
• 제17C권2호
• /
• pp.145-152
• /
• 2010

본 논문에서는 표준기저(standard basis) 표기법을 이용하여 GF($2^{163}$) 상에서개선된 나눗셈 알고리듬을 제안하고, 제안한 알고리듬을 기반으로 한 반복 하드웨어 구조(iterative hardware structure)를 갖는 고속 나눗셈기를 설계한다. 제안한알고리듬은 이진 확장 GCD 알고리듬을 기본으로 하고 있으며, 모듈러감소 (modular reduction)를 위한 모든 산술연산은 기존의 방법과 달리 하나의 while루프 내에서 수행된다. 제안된 알고리듬을 기본으로 하여 설계된 나눗셈기는 모듈러 연산을 위한 각 모듈이 하나의 클럭에 의해서제어되므로 계산 속도가 매우 빠르다. 여기에서 사용하는 감소 다항식(reduction polynomial)은 SEC2 (Standards for Efficient Cryptography) 에서 권장하는 $f(x)=x^{163}+x^7+x^6+x^3+1$이며, 차수(degree) m은 163을 사용한다. 제안한 알고리듬은 Verilog HDL(Hardware Description Language)을 사용하여 FPGA로 구현되었으며, Xilinx-VirtexII XC2V8000 FPGA 상에서 85MHz로 동작함을 확인하였다. 또한, 구현 결과 및 성능 평가를 통하여 제안한 알고리듬의 종래의 두 알고리듬보다 성능이크게 개선됨을 보인다.

• 정보보호학회논문지
• /
• 제22권3호
• /
• pp.439-445
• /
• 2012

오류 주입 공격은 비밀 키를 내장하여 사용하는 암호 장치에서 연산 수행시 공격자가 오류를 주입하는 방법으로 비밀 키를 찾아낼 수 있어 암호시스템 운영의 심각한 위협이 되고 있다. 논문에서는 AES 암호 연산을 수행하는 동안 라운드 함수를 반복적으로 사용하는 경우, 반복하는 구문에 오류를 넣어 한 라운드를 생략하면 쉽게 비밀 키를 추출할 수 있음을 보이고자 한다. 제안하는 공격 방법을 상용 마이크로프로세서에서 실험한 결과, 두 개의 정상-오류 암호문 쌍만 있으면 128비트 AES 비밀 키가 노출됨을 확인하였다.

• Journal of applied mathematics & informatics
• /
• 제5권1호
• /
• pp.73-90
• /
• 1998

In this paper we introduce and study a number of new classes of quasi variational inequalities. using essentially the projection technique and its variant forms we prove that the gen-eralized set-valued mixed quasivariational inequalities are equivalent to the fixed point problem and the Wiener-Hopf equations(normal maps). This equivalence enables us to suggest a number of iterative algorithms solving the generalized variational inequalities. As a special case of the generalized set-valued mixed quasi variational in-equalities we obtain a class of quasi variational inequalities studied by Siddiqi Husain and Kazmi [35] but there are several inaccuracies in their formulation of the problem the statement and the proofs of the problem the statement and the proofs of their results. We have removed these inaccuracies. The correct formulation of thir results can be obtained as special cases from our main results.

• 대한수학회지
• /
• 제59권5호
• /
• pp.1015-1045
• /
• 2022

Zagier introduced the term "strange identity" to describe an asymptotic relation between a certain q-hypergeometric series and a partial theta function at roots of unity. We show that behind Zagier's strange identity lies a statement about Bailey pairs. Using the iterative machinery of Bailey pairs then leads to many families of multisum strange identities, including Hikami's generalization of Zagier's identity.

• 한국국방경영분석학회지
• /
• 제14권1호
• /
• pp.42-52
• /
• 1988

Many programmers start debugging by reading the faulty program from start to bottom without investigating carefully the erroneous program. Expert programmers, however, trace backward from a particular variable in a specific statement to identify all possible sources of influence on the value of variable (program slice). Weiser proposed a slicing algorithm (method) that is complex, iterative and still in modification [3,4]. This paper presents a method to generate a program slice by use of matrix computation which represents all possible slices of the program. The matrix representation of a program is soundly based on the graph theory of data dependency.

• 한국국방경영분석학회지
• /
• 제13권2호
• /
• pp.42-52
• /
• 1987

Many programmers start debugging by reading the faulty program from start to bottom without investigating carefully the erroneous program. Expert programmers, however, trace backward from a particular variable in a specific statement to identify all possible sources of influence on the value of variable (program slice). Weiser proposed a slicing algorithm (method) that is complex, iterative and still in modification [3,4]. This paper presents a method to generate a program slice by use of matrix computation which represents all possible slices of the program. The matrix representation of a program is soundly based on the graph theory of data dependency.

• Structural Engineering and Mechanics
• /
• 제45권2호
• /
• pp.169-182
• /
• 2013

The mechanical behavior of rectangular foundation plates with perimetric beams and internal stiffening beams of the plate is herein analyzed, taking the foundation design into account. A series of dimensionless parameters related to the geometry of the studied elements were defined. In order to generalize the problem statement, an initial settlements was considered. A numeric procedure was developed for the resolution by means of the Finite Differences Method that takes into account the stiffness of the plate, the perimetric and internal plate beams and the soil reaction module. Iterative algorithms were employed which, for each of the analyzed cases, made it possible to find displacements and reaction percentages taken by the plate and those that discharge directly into the perimetric beams, practically without affecting the plate. To enhance its mechanical behavior the internal stiffening beams were prestressed and the results obtained with and without prestressing were compared. This analysis was made considering the load conditions and the soil reaction module constant.

• Structural Engineering and Mechanics
• /
• 제52권5호
• /
• pp.1051-1067
• /
• 2014

The structures of concern in this study are subject to two types of forces: dead loads from the acceleration imposed on the structures as well as the installed operation machines and the additional adjustable forces. We wish to determine the critical values of the adjustable forces when buckling of the structures occurs. The mathematical statement of such a problem gives rise to a constrained eigenvalue problem (CEVP) in which the dominant eigenvalue is subject to an equality constraint. A numerical algorithm for solving the CEVP is proposed in which an iterative method is employed to identify an interval embracing the target eigenvalue. The algorithm is applied to four engineering application examples finding the critical loads of a fixed-free beam subject to its own body force, two plane structures and one wide-flange beam using shell elements when acceleration force is present. The accuracy is demonstrated using the first example whose classical solution exists. The significance of the equality constraint in the EVP is shown by comparing the solutions without the constraint on the eigenvalue. Effectiveness and accuracy of the numerical algorithm are presented.