• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.05a
• /
• pp.145-148
• /
• 2018

클라우드 컴퓨팅 기반 사물인터넷 환경은 급격히 증가하는 통신량, 기종 간 이질성, 지연 시간과 같은 문제점으로 인해 어려움을 겪고 있다. 이를 해결하기 위한 대표적인 방법 중 하나는 분산 모델을 통해 클라우드 컴퓨팅 환경에 집중된 네트워크 또는 컴퓨팅 파워를 분산시키는 포그 컴퓨팅 (Fog Computing) 또는 에지 컴퓨팅 (Edge Computing)을 활용하는 것이다. 그러나 이 분산형 네트워크의 단점을 보완하기 위해 사물인터넷 (IoT, Internet of Things)과 가장 가까이 존재하는 네트워크 모델로써 미스트 컴퓨팅 (Mist Computing)이 탄생하였다. 그러나 다양한 프로토콜에 의해 통신이 이루어지는 사물인터넷 환경에는 수천 가지 제로데이 공격이 존재한다. 이 공격들의 대부분은 이전에 알려진 공격의 작은 변형체이다. 이러한 공격을 효과적으로 막기 위해 사물인터넷 환경에서의 침입 탐지 시스템은 지능적이어야 한다. 따라서 본 논문에서는, 미스트 컴퓨팅 환경에서 새로운 또는 지속적으로 변화하는 사물인터넷 대상 공격을 효과적으로 방어하기 위한 인공지능 기반 침입 탐지 시스템을 제안한다.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2018.05a
• /
• pp.132-132
• /
• 2018

홍수피해 발생 시 각 지역의 특성에 따라 피해의 정도가 다르게 나타나고 있다. 도시지역의 경우 포장도로로 인하여 불투수 면적이 넓어 배제 시설의 허용 용량을 초과하는 강우 발생 시 지표면 위로 유량이 누적된다. 특히 농촌지역에 비해 도시지역에서 국지적으로 기습적인 강우에 빈번히 침수피해가 발생한다. 우리나라 도시지역은 인구와 건물이 밀집되어 재산과 인명피해 및 교통체증 등으로 인한 간접피해 등 농촌지역에 비해 더 큰 피해가 발생할 수 있다. 도시 지역의 침수 피해 저감을 위해 우수 유출 저감시설과 내수배제 시설을 설치하여 우수의 유출을 지연시키거나 인위적으로 방류부 하천으로 방류시키는 방법이 있지만 이러한 시설의 설치에는 막대한 비용이 필요하며, 내수배제시설이 설치된 장소에 따라 침수현상이 다르게 나타나 저감시설 설치 전 공간적, 효율적으로 설치 가능한 장소와 침수 저감 효율에 대한 분석이 필요하다. 본 연구는 도림1 배수분구를 대상으로 XP-SWMM모델을 이용하였으며, 2010년 09월 21일 침수 발생 당시의 강우를 적용하여, 연구대상지역의 도시유출해석을 진행하였다. 또한, 어린이 공원과 학교운동장에 가상지하저류조를 설치하였을 시 침수저감 효과에 대하여 연구를 진행하였으며, 다양한 지하저류조 설치 시나리오에 따른 침수저감효과를 수심별 침수면적으로 나타내어 분석을 실시하였다. 본 연구를 통해, 도심지 지하 저류시설 도입의 기초자료로 활용될 수 있을 것으로 판단된다.

• The Journal of Korean Institute of Next Generation Computing
• /
• v.15 no.3
• /
• pp.16-24
• /
• 2019

Recently, research on emotion recognition based on EEG has attracted great interest from human-robot interaction field. In this paper, we propose a method of labeling using image-based EEG topography instead of evaluating emotions through self-assessment and annotation labeling methods used in MAHNOB HCI. The proposed method evaluates the emotion by machine learning model that learned EEG signal transformed into topographical image. In the experiments using MAHNOB-HCI database, we compared the performance of training EEG topography labeling models of SVM and kNN. The accuracy of the proposed method was 54.2% in SVM and 57.7% in kNN.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.1
• /
• pp.61-67
• /
• 2022

In this paper, we propose a model that can convert some of the simulation program resources to a mobile environment. Recently, smart factories that use PLCs as controllers in the manufacturing industry are rapidly becoming widespread. However, in the situation where it is difficult to operate due to the shortage of PLC operation personnel, the actual situation is that a platform for PLC operation education is necessary. Currently most PLC-related educational platforms are based on 2D, which makes accurate learning difficult and difficult. When a simulation program is applied to distance learning in a general PC environment, many elements are displayed on the monitor, which makes screen switching inconvenient. Experiments with the proposed model confirmed that there was no frame deterioration under general circumstances. The average response time by the request frame was 102 ms, and it was judged that the learner was not at the level of experiencing the system delay.

• Journal of Platform Technology
• /
• v.11 no.4
• /
• pp.71-79
• /
• 2023

In this paper, we analyze the characteristics of machine learning workloads and, based on them, propose a distributed in-memory caching technique to improve the performance of machine learning workloads. The core of machine learning workload is model training, and model training is a computationally intensive task. Performing machine learning workloads in a Kubernetes-based cloud environment in which the computing framework and storage are separated can effectively allocate resources, but delays can occur because IO must be performed through network communication. In this paper, we propose a distributed in-memory caching technique to improve the performance of machine learning workloads performed in such an environment. In particular, we propose a new method of precaching data required for machine learning workloads into the distributed in-memory cache by considering Kubflow pipelines, a Kubernetes-based machine learning pipeline management tool.

• Journal of Radiation Protection and Research
• /
• v.32 no.4
• /
• pp.178-183
• /
• 2007

Neutrons are high LET (linear energy transfer) radiation and cause more damage to the target cells than x-rays or gamma rays. The damage from neutrons is generally considered fatal to a cell and neutrons have a greater tendency to cause cell death through direct interaction on DNA. We performed experiments to measure growth delay ratio and oxygen enhancement ratio (OER) in mouse model system. We inoculated EMT-6 cells to the right hind leg of BALB-c mouse and X-rays and neutron beams were given when the average volume of tumors reached $200-300mm^3$. We irradiated 0, 11, 15.4 Gy of X-ray and 0, 5, 7 Gy of fast neutron beam at normoxic and hypoxic condition. The volume of tumors was measured 3 times per week. In x-ray experiment, growth delay ratio was 1.34 with 11 Gy and 1.33 with 15.4 Gy in normoxic condition compared to in hypoxic condition, respectively. In neutron experiment, growth delay ratio was 0.94 with 5 Gy and 0.98 with 7 Gy, respectively. The OER of neutron beam was 0.97. The neutron beam was more effective than X-ray in the control of hypoxic tumors.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.6
• /
• pp.646-652
• /
• 2017

The objective of this study was to verify the longitudinal reciprocal causal relationship between depression and the somatic symptoms of depression among adolescents through an autoregressive cross-lagged model using data from the Korean Children & Youth Panel Survey. The subjects of this study included 1,968 adolescents, who participated in the second, fourth, and sixth Korean Children & Youth Panel Surveys. The results showed that both depression and the somatic symptoms of depression at a previous point in time affect depression and the somatic symptoms of depression at a later point in time. It was also found that depression at a previous point in time has a cross-lagged effect on the somatic symptoms of depression at a later point in time, implying that more severe depression at a previous point in time leads to increased severity of somatic symptoms at a later point in time. It was found that the somatic symptoms of depression at a previous point in time have a cross-lagged effect on depression at a later point in time, indicating that more severe somatic symptoms of depression at a previous point in time lead to increased severity of depression at a later point in time. This study is significant in that it provides baseline information about nursing interventions for adolescent mental health.

• Journal of the Korean Society of Groundwater Environment
• /
• v.5 no.3
• /
• pp.155-161
• /
• 1998

Retardation effect of heavy metals in soils caused by adsorption onto the surfaces of solids particles is well known phenomenon. In this study, we investigated the retardation effect on the mobility of a Zn in a sandy soil by conducting batch and column tests. The column test consisted of monitoring the concentrations of effluent versus time known as a breakthrough curve (BTC). We used NaCl and ZnCl$_2$ solutions with the concentration of 10 g/L as a tracer, and injected them respectively into the inlet boundary of the soil sample as a square pulse type, and monitored the effluent concentrations at the exit boundary under a steady state condition using an EC-meter and ICP-AES. The batch test was conducted based on the standard procedure of equilibrating fine fractions collected from the soil with various initial ZnCl$_2$ concentrations, and analysis of Zn ions in the equilibrated solutions using ICP-AES. The results of column test showed that i) the peak concentration of ZnCl$_2$analyzed by ICP was far less than that of either NaCl or bulk electrical conductivity and ⅱ) travel times of peak concentrations for two tracers were more less identical. The relatively low concentration of Zn can be explained by ion exchange between Zn and other cations, and possible precipitation of Zn in the form of Zn(OH)$_2$due to high pH range (7.0∼7.9) of the effluent. The identical result of travel times of peak concentrations indicates that the retardation effect is not present in the soil. The only way to describe the prominent decrease of Zn ion was to introduce decay or sink coefficient in the CDE model to account for irreversible decrease of Zn ions in the aqueous phase.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.44 no.1
• /
• pp.33-39
• /
• 2007

LonWorks over IP (LonWorks/IP) virtual device network (VDN) is an integrated form of LonWorks device network and IP data network. LonWorks/IP VDN can offer ubiquitous access to the information on the factory floor and make it possible for the predictive and preventive maintenance on the factory floor. Timely response is inevitable for predictive and preventive maintenance on the factory floor under the real-time distributed control. The network induced uncertain time delay deteriorates the performance and stability of the real-time distributed control system on LonWorks/IP virtual device network. Therefore, in order to guarantee the stability and to improve the performance of the networked distributed control system the time-varying uncertain time delay needs to be compensated for. In this paper, under the real-time distributed control on LonWorks/IP VDN with uncertain time delay, a control scheme based on disturbance observer and ZPETC(Zero Phase Error Tracking Controller) phase lag compensator is proposed and tested through computer simulation. The result of the proposed control is compared with that of internal model controller (IMC) based on Smith predictor and disturbance observer. It is shown that the proposed control scheme is disturbance and noise tolerant and can significantly improve the stability and the tracking performance of the periodic reference. Therefore, the proposed control scheme is well suited for the distributed servo control for predictive maintenance on LonWorks/IP-based virtual device network with time-varying delay.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.20 no.5
• /
• pp.809-822
• /
• 2018

A quantitative risk assessment has been introduced to quantitatively evaluate fire risk as a means of performance based fire protection design in the design of railway tunnel disaster prevention facilities. However, there are insufficient studies to examine the effect of various risk factors on the risk. Therefore, in this study, the risk assessment was conducted on the model tunnel in order to examine the effects of the evacuation start time delay and the fire growth curve on the quantitative risk assessment. As a result of the analysis of the scenario, the fatalities occurred mainly when escapes in the same direction as the direction of the fire smoke movement. In addition, after the FED exceeded 0.3, the maximum fatalities occurred within 10 minutes. In the range of relatively low risk, distance between cross passages, evacuation delay time and fire growth curve were found to affect the risk, but they were found to have little effect on the condition that the risk reached the limit. Especially, in this study, it was evaluated that the evacuation delay time reduction, fire intensity and duration reduction effect were not observed when the distance between cross passages was more than 1500 m.