• 정보처리학회논문지A
• /
• 제14A권2호
• /
• pp.99-106
• /
• 2007

지역적으로 분산되어 있는 이기종의 시스템들을 하나로 묶어 사영하는 그리드 컴퓨팅이 차세대 병렬.분산 연산을 위한 새로운 패러다임으로 관심을 끌고 있다. 고속 네트워크로 연결된 다수의 컴퓨터 시스템이 사용자에게 통합된 가상의 컴퓨팅 서비스를 제공하는 그리드 시스템은 통신비용에 대한 중요성이 매우 크다. 따라서 그리드 환경에서 스케줄링 알고리즘은 작업의 실행시간을 단축하기 위하여 자원들의 연산능력과 함께 통신에 대한 비용을 고려하여야 한다. 그러나 현재까지 발표된 대부분의 스케줄링 알고리듬들은 작업이 한 클러스터에서 처리되는 것을 가정함으로써 통신비용을 무시하였으며, 작업이 다수의 클러스터에 분산되어 처리되는 경우에도 통신비용에 관한 오버헤드를 고려하지 않았다. 본 논문에서는 그리드 시스템에 적합한 기존 스케줄링 알고리즘들의 성능을 분석하였으며, 작업이 다수의 클러스터에 분산되어 수행되는 co-allocation 환경에서 통신비용을 고려하여 알고리즘들의 성능을 비교하고 분석하였다.

• 한국콘텐츠학회논문지
• /
• 제14권5호
• /
• pp.8-17
• /
• 2014

슈퍼컴퓨팅 자원들은 주로 MPI와 같은 메시지 교환 인터페이스에 기반한 통신 집적도가 높은 고성능 컴퓨팅(HPC: High Performance Computing) 응용 분야를 지원하는데 활용되어 왔다. 반면에, 대규모 계산처리 컴퓨팅(HTC: High Throughput Computing) 방식의 패러다임은 주로 계산 집적도가 높고(상대적으로 적은 I/O 연산), 독립적인(작업들 간의 통신이 적음) 많은 수의 작업을 처리하는 것을 요구하고 있다. 국내에서도 고에너지 물리, 신약개발, 핵물리와 같은 연구 분야를 중심으로 대규모 컴퓨팅 자원을 요구하는 계산처리에 대한 수요가 증가하고 있다. 본 논문에서는 이러한 HTC 과학 응용들에 대한 효율적인 지원을 국가차원의 슈퍼컴퓨팅 분산 환경에서 제공하기 위해 연구/개발되어진 대규모 계산처리 서비스(HTCaaS: High Throughput Computing as a Service)의 전체 구조 및 구성 요소, 실행 시나리오 및 실제 응용 적용 사례 등에 대해 서술한다.

• Transactions on Electrical and Electronic Materials
• /
• 제15권2호
• /
• pp.103-111
• /
• 2014

This paper investigates the output powers of PV modules by predicting three unknown parameters: reverse saturation current, and series and shunt resistances. A theoretical model using the non-uniform physical parameters of solar cells, including the temperature coefficients, voltage, current, series and shunt resistances, is proposed to obtain the I-V characteristics of PV modules. The solar irradiation effect is included in the model to improve the accuracy of the output power. Analytical and Newton methods are implemented in MATLAB to calculate a module output. Experimental data of the non-uniform solar cells for both serial and parallel connections are used to extend the implementation of the model based on the I-V equation of the equivalent circuit of the cells and to extend the application of the model to m by n modules configuration. Moreover, the theoretical model incorporates, for the first time, the variations of series and shunt resistances, reverse saturation current and irradiation for easy implementation in real power generation. Finally, this model can be useful in predicting the degradation of a PV system because of evaluating the variations of series and shunt resistances, which are critical in the reliability analysis of PV power generation.

• 전기학회논문지
• /
• 제63권3호
• /
• pp.358-366
• /
• 2014

This paper proposes a quasi Z-source AC-AC converter with the low DC voltage distribution capability operating as a power electronic transformer. The proposed system has configuration that the input terminals of two quasi Z-source AC-AC converters are connected in parallel, also their output terminal are connected in series. Simple control method of duty ratio was proposed for the in phase buck-boost AC voltage mode and the DC output voltage control. DSP based experiment and PSIM simulation were performed. As a result, the PSIM simulation results were same with the measured results. By controlling the duty ratio under the condition of 100 [${\Omega}$] load, quasi Z-source AC-AC converter could buck and boost the AC output voltage in phase with the AC input voltage, and the same time, the constant DC voltage could be output without affecting the AC output characteristics. And, the DC output voltage 48[V] was constantly controlled in dynamic state in case while the load is suddenly changed ($50[\Omega]{\rightarrow}100[\Omega]$). From the above result, we could know that the quasi Z-source AC-AC converter can act as a power electronic transformer with a low DC voltage distribution capability.

• Toxicological Research
• /
• 제17권2호
• /
• pp.73-82
• /
• 2001

The incidence and distribution of necrotic and apoptotic neural cells, and activated astrocytes in the brain of rats intoxicated intra peritoneally with diisopropylfluorophosphate were investigated. Pyridostigmine bromide (0.1 mg/kg) and atropine methylnitrate (20 mg/kg) were pretreated intramuscularly 30 min and 10 min, respectively, prior to diisopropylfluorophosphate (4-10 mg/kg) administration. Diisopropylfluorophosphate induced severe limbic seizures, early necrotic and delayed apoptotic brain injuries, and rapid astrocytic responses. The necrosis, which was closely related to seizure intensity, was observed as early as 1 hr after intoxication predominently in hippocampal pyramidal cells, cerebellar Purkinje cells and neurons in pyriform/entorhinal cortices, showing malacia of neurophils. In contrast, apoptosis started to appear 12 hr after intoxication in neurons in thalamus, amygdala and neocortex, and ephendymal cells surrounding the 4th ventricle. Since marked apoptosis was induced in rats exhibiting relatively-low seizure intensity, the degree of necrosis and apoptosis was shifted to each type of injury according to the seizure intensity. Activated astrocytes, observed within 1 hr along the limbic system, were suggested to affect the neural injury patterns by producing high level of nitric oxide. However, the distribution of activated astrocytes was not in parallel with those of necrotic or apoptotic injuries, implying that the astrocytic responses resulted from seizure activity rather than neural injuries. Furthermore, astrocytes in malacic tissues disappeared during the severe limbic seizures. Therefore, it would be one of the cautionary notes on the expression of glial fibrillary acidic protein in astrocytes as a biochemical marker of brain injuries following acute exposure to organophosphates.

• 자원환경지질
• /
• 제5권3호
• /
• pp.133-144
• /
• 1972

The mine of our present concern is situated at Shim-ri, Gusan-myon, Changwon-gun, Kyongsang-namdo, with lattitude $128^{\circ}35^{\prime}{\sim}36^{\prime}N$ and longitude $35^{\circ}03^{\prime}{\sim}04^{\prime}E$. This mine has not been noticed until the intermittent geological survey for the ore deposits were initiated from September, 1967 till 1970. The main mineralized zones, No.1 and No.2 zones, were studied by the diamond drilling of 9 holes down to the total depth of 1,140m, and found to have ore reserves of Cu 1.99% ore, estimated to reach around $358,000{\frac{M}{T}}$ (proved $117,000{\frac{M}{T}}$, indicated $241,000{\frac{M}{T}}$), which triggered the new exploitation of this mine. Geological composition of the district near the mine is mainly from the andesite belonging to the Silla Series of Kyongsang System and the distribution is broadly spread. Ore deposits are the hydrothermal one, filling the shear zone formed alongside the andesite main joint. There are two stripes of copper bearing mineralized zone which are about 40~70 meters apart and parallel to each other, in addition to which two others are expected. The strike of the main mineralized zone lies at $N15^{\circ}{\sim}20^{\circ}W$, the dip at $60^{\circ}{\sim}70^{\circ}NE$. The principal components of the ore mineral are chalcopyrite, bornite and as secondary, cuprite, tenorite, azurite and malachite. Pyrite, magnetite, specular hematite, very little of galena, sphalerite, quartz, epidote, zoisite, chlorite and calcite are found as gangue mineral. Confirmed by the drilling, the main mineralized zone, No.1 zone, has the length of 320m, average width of 1.62m, Ag 26g/t, Cu 1.84% and the average width of the ore zone except the narrow barren andesite horse interposed in the mineralized zone is 1.32m with Ag 32g/t and Cu 2.26%. The mineralized zone No.2 is 340m long, 1.00m wide with Ag 30g/t and Cu 2.15%. Neglecting the barren andesite horse the width is 0.73m with Ag 42g/t and Cu 2.94%.

• 방송공학회논문지
• /
• 제10권1호
• /
• pp.83-102
• /
• 2005

The problem of video scheduling is analyzed in the framework of divisible load scheduling. A divisible load can be divided into any number of fractions (parts) and can be processed/computed independently on the processors in a distributed computing system/network, as there are no precedence relationships. In the video scheduling, a frame can be split into any number of fractions (tiles) and can be processed independently on the processors in the network, and then the results are collected to recompose the single processed frame. The divisible load arrives at one of the processors in the network (root processor) and the results of the computation are collected and stored in the same processor. In this problem communication delay plays an important role. Communication delay is the time to send/distribute the load fractions to other processors in the network. and the time to collect the results of computation from other processors by the root processors. The objective in this scheduling problem is that of obtaining the load fractions assigned to each processor in the network such that the processing time of the entire load is a minimum. We derive closed-form expression for the processing time by taking Into consideration the communication delay in the load distribution process and the communication delay In the result collection process. Using this closed-form expression, we also obtain the optimal number of processors that are required to solve this scheduling problem. This scheduling problem is formulated as a linear pro-gramming problem and its solution using neural network is also presented. Numerical examples are presented for ease of understanding.

• Smart Structures and Systems
• /
• 제14권5호
• /
• pp.901-916
• /
• 2014

Magnetorheological (MR) fluid is one of the field-responsive fluids that is of interest to many researchers due to its high yield stress value, which depends on the magnetic field strength. Similar to electrorheological (ER) fluid, the combination of working modes is one of the techniques to increase the performance of the fluids with limited focus on MR fluids. In this paper, a novel MR testing cell incorporated with valve, shear and squeeze operational modes is designed and constructed in order to investigate the behaviour of MR fluid in combined mode. The magnetic field distribution in the design concept was analyzed using finite element method in order to verify the effective areas of each mode have the acceptable range of flux density. The annular gap of valve and shear were fixed at 1 mm, while the squeeze gap between the parallel circular surfaces was varied up to 20 mm. Three different coil configurations, which were made up from 23 SWG copper wires were set up in the MR cell. The simulation results indicated that the magnetic field distributed in the squeeze gap was the highest among the other gaps with all coils were subjected to a constant applied current of 1 A. Moreover, the magnetic flux densities in all gaps were in a good range of magnitude based on the simulations that validated the proposed design concept. Hence, the 3D model of the MR testing cell was designed using Solidworks for manufacturing processes.

• 정보과학회 컴퓨팅의 실제 논문지
• /
• 제23권3호
• /
• pp.188-193
• /
• 2017

영상 장비가 발전하고 고화질 이미지의 사용량이 많아짐에 따라 이를 병렬화하여 빠르게 처리하는 방법이 많이 연구되고 있다. 본 논문은 OpenCL을 사용한 이미지 처리 프로그램이 하드웨어에 관계없이 최적의 성능을 내도록 하는 방법을 제시한다. OpenCL은 계산 디바이스의 아키텍처에 따라 적절한 병렬화 방법과 워크그룹 크기를 개발자가 실험을 통해 찾아야 한다. 하지만 이는 시간적, 자원적 한계가 있다. 본 논문에서는 대상 하드웨어에 최적화된 병렬화 방법과 워크그룹 크기를 프로파일링을 통해 자동으로 선택하는 방법을 제안한다. 또한 OpenCL을 사용한 이미지 처리는 입력 이미지의 크기에 따라 순차코드 보다 성능이 좋을 수도 있고 안좋을 수도 있다. 본 논문에서는 OpenCL코드의 성능이 순차코드 성능보다 좋아지는 시점의 이미지 크기를 자동으로 검색하여 OpenCL을 사용할지 기존의 순차코드를 사용할지 결정하는 이미지 크기 threshold를 설정하는 방법을 제시한다. 이를 통해 이미지 처리 프로그램이 임의의 하드웨어에서 이미지 크기에 상관없이 항상 최적의 성능을 보이는 것을 확인했다.

• 정보과학회 컴퓨팅의 실제 논문지
• /
• 제21권12호
• /
• pp.745-750
• /
• 2015

LiDAR 기술은 실제와 유사한 수준의 3차원 영상 정보를 제공하는 센싱 기술로, 다양한 분야에서 활용되고 있다. 이 기술을 실제 활용하기 위해서는 복잡하고, 방대한 데이터 처리가 요구되며, 이에 급변하는 상황에서 빠른 대응이 필요한 군사 작전에서의 활용은 제한적이다. 본 논문에서는 무인기 플랫폼 기반의 실시간 LiDAR 데이터 처리구조를 제안하여 LiDAR 데이터 처리를 위해 무인기-고성능 데이터 처리영역 간 병렬처리와 오프로딩을 통해 실시간 수준의 탐지 정보를 제공하도록 제안한다. 또한, 제안 구조의 적용 가능성 검증을 위해 기존 처리 방식과 무인기 플랫폼의 처리환경과 유사한 ARM 클러스터에서의 병렬 처리 방식에서 실제 LiDAR 데이터의 보간 처리를 통해 처리시간, 용량을 확인하는 실험을 수행하였고, 기존 방식과 유사하거나 양호한 결과를 확인하였다. 향후 제안 구조를 기반으로 실시간성 보장이 요구되는 무인 침투자 감시 등 군사 작전에 활용될 수 있을 것이다.