• Proceedings of the Korean Society of Computer Information Conference
• /
• 2016.07a
• /
• pp.13-15
• /
• 2016

에너지 소비가 큰 건물은 내부 온/습도, 이산화탄소 농도, 미세먼지 농도 등의 일정 공기 질을 유지하면서 에너지 비용을 최소화할 수 있는 제어계획을 수립하는 것이 필요하다. 기존 건물에서 실내 환경의 운영은 설정된 실내 환경 값을 기준을 벗어나면 설비 기기를 제어하는 방식으로 이루어진다. 이는 단 시간에 고에너지를 투입하여 장비를 가동시키므로 에너지 소모가 크며 peak 전력이 높아 에너지 비용이 크다는 문제가 있다. 따라서 온도를 포함한 환경이 변해가는 상황을 예측하고 사전에 에너지 사용 계획을 수립하여 관리 제어를 수행함으로써 예열부하 등의 불필요한 에너지 손실을 절감하려 한다. 이를 위해 실내 환경이 변화하는 것을 예측하고 후보 제어계획으로 제어를 수행할 때 소요되는 에너지가 어느 정도인지 시뮬레이션하여 제어계획의 적합도를 평가한다. 기존 EnergyPlus와 같은 시뮬레이션 도구는 모델이 복잡하여 시뮬레이션에 많은 시간이 필요하기 때문에 환경 변화를 반영하기 위해 주기적으로 재수립되는 수많은 제어계획 데이터를 단시간에 시뮬레이션하기에 부적합하다. 본 논문에서는 빠른 시뮬레이션을 위해 실제 운영 데이터와 에뮬레이션을 통해 획득한 운영 데이터를 기반으로 학습 알고리즘을 이용하여 제어계획 적용 시의 미래 상황을 예측한다.

• Journal of Convergence for Information Technology
• /
• v.10 no.6
• /
• pp.41-48
• /
• 2020

The sensor nodes in underwater IoT networks have practical limitations in power supply. Thus, the reduction of power consumption is one of the most important issues in underwater environments. In this regard, AMC(Adaptive Modulation and Coding) techniques are used by using the relation between SNR and BER. However, according to our hands-on experience, we observed that the relation between SNR and BER is not that tight in underwater environments. Therefore, we propose a deep learning based MLP classification model to reflect multiple underwater channel parameters at the same time. It correctly predicts BER with a high accuracy of 85.2%. The proposed model can choose the best parameters to have the highest throughput. Simulation results show that the throughput can be enhanced by 4.4 times higher than the conventionally measured results.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.26 no.5C
• /
• pp.291-301
• /
• 2006

In this study the characteristics of electrokinetic remediation, which is dependent on a various electrode configuration, was predicted from 2-D numerical analysis program (HERO-2D). Based on the predicted results for one dimensional and two dimensional electrode configurations, the optimized electrode configuration was determined by analyzing remediation efficiency, consumed electric power, installation cost of electrode and so on. When proposed electrode configurations were applied for in-situ remediation of the soils contaminated by heavy metals, the electrode configuration of high remediation efficiency should be chosen in case the high removal effect would be required, and one dimensional electrode configuration should be chosen in case the hard field works would be expected. Because the rectangular electrode configuration is better than others for consumed electric power, remediation efficiency per unit power, installation cost of electrode and so on, it can obtain the best results for the cost reduction.

• Journal of the Korean housing association
• /
• v.23 no.6
• /
• pp.31-38
• /
• 2012

In this paper, we draws tendency of the electricity consumption in residential buildings according to inhabitants Composition types and the level of incomes. it is necessary to reduce energy cost and keep energy security through the electricity demand forecasting and management technology. Progressive social change such as increases of single household, the aging of society, increases in the income level will replace the existing residential electricity demand pattern. However, Only with conventional methods that using only the energy consumption per-unit area are based on Energy final consumption data can not respond to those social and environmental change. To develop electricity demand estimation model that can cope flexibly to changes in the social and environmental, In this paper researches propensity of electricity consumption according to the type of residents configuration, the level of income. First, we typed form of inhabitants in residential that existed in Korea. after that we calculated hourly electricity consumption for each type through National Time-Use Survey performed at the National Statistical Office with considering overlapping behavior. Household appliances and retention standards according to income level is also considered.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.05a
• /
• pp.458-460
• /
• 2019

최근 클라우드 시스템 환경이 점차 늘어남에 따라 데이터 센터(IDC) 구축이 점차 늘어나가고 있다. 데이터 센터는 최근 부각하고 있는 4 차 산업 영역에서 사물 인터넷(IoT), 자율주행차 등 에서 처리될 대용량 데이터로 인한 이를 처리하는 중요한 역할을 담당하고 있다. 데이터센터 운영에는 대량의 에너지가 필요하다. 수 많은 컴퓨터에서 발생하는 열에너지를 처리하기 위하여 대량의 전력 냉방 에너지를 소비하고 있다. 냉방 공조 운영은 데이터 센터 운영에 중요한 역할을 한다. 이유는 많은 컴퓨터를 가동하는 비용보다 부대 시설로 운영되는 냉방 에너지를 보다 많이 소비하는 현상까지 발생하고 있다. 이에 최근 데이터 센터 냉방 공조 운영을 효율화하는 것에 연구를 맞추고 있다. 본 논문에서는 냉방 공조 운영 효율화 하도록 하기 위해서 다중 기계 학습을 활용한 데이터 센터의 냉방 에너지 절감 시스템을 제안하고자 한다. 기존의 단수 알고리즘을 활용하여 머신 러닝의 모델구현 방식이 아닌 다중의 기계 학습을 통하여 최적화된 모델을 일일 배치로 생성하여 예측을 하는 시스템이다. 본 시스템을 통하여 사전에 최적화된 냉방 운영을 하여 기존 데이터 센터의 운영되는 과다 냉방을 감축 시켜 에너지를 절감해주는 기능을 제공한다. 본 논문 시스템 연구 결과는 폭발적으로 늘어가고 있는 데이터 센터의 에너지 효율화에 기여할 수 있고, 클라우드 사업에서 경쟁력을 줄 수 있는 운영 시스템 방안을 제시한다.

• Journal of IKEEE
• /
• v.25 no.2
• /
• pp.388-398
• /
• 2021

This paper proposes a particle swarm optimization (PSO)-based peak shaving scheme using energy storage system (ESS) for electricity tariff reduction. The proposed scheme compares the actual load with the estimated load consumption, calculates the additional output power that the ESS needs to discharge additionally to reduce peak load, and adds the input. In addition, in order to compensate for the additional power, the process of allocating power to the determined point is performed, and an optimization that minimizes the average of the load expected at the active power allocations using PSO so that the allocated value does not affect the peak load. To investigated the performance of the proposed scheme, case study of small and large load prediction errors was conducted by reflecting actual load data and load prediction algorithm. As a result, when the proposed scheme is performed with the ESS charge and discharge control to reduce electricity tariff, even when the load prediction error is large, the peak load is successfully reduced, and the peak load reduction effect of 17.8% and electricity tariff reduction effect of 6.02% is shown.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.47 no.5
• /
• pp.153-164
• /
• 2010

A program consumes energy by executing its instructions. The amount of cosumed power is mainly proportional to algorithm complexity and it can be calculated by using complexity information. Generally, the complexity of a S/W is estimated by the microprocessor simulator. But, the simulation takes long time why the simulator is a software modeled the hardware and it only provides the information about computational complexity quantitatively. In this paper, we propose a complexity estimation method of analysis of S/W on source code level and produce the complexity metric mathematically. The function-wise complexity metrics give the detailed information about the calculation-concentrated location in function. The performance of the proposed method is compared with the result of the gate-level microprocessor simulator 'SimpleScalar'. The used softwares for performance test are $4{\times}4$ integer transform, intra-prediction and motion estimation in the latest video codec, H.264/AVC. The number of executed instructions are used to estimate quantitatively and it appears about 11.6%, 9.6% and 3.5% of error respectively in contradistinction to the result of SimpleScalar.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.30 no.6
• /
• pp.1291-1300
• /
• 2020

Recently, a deep learning-based non-profiling side-channel analysis was proposed. The deep learning-based non-profiling analysis is a technique that trains a neural network model for all guessed keys and then finds the correct secret key through the difference in the training metrics. As the performance of non-profiling analysis varies greatly depending on the neural network training model design, a correct model design criterion is required. This paper describes the two types of loss functions and eight labeling methods used in the training model design. It predicts the analysis performance of each labeling method in terms of non-profiling analysis and power consumption model. Considering the characteristics of non-profiling analysis and the HW (Hamming Weight) power consumption model is assumed, we predict that the learning model applying the HW label without One-hot encoding and the Correlation Optimization (CO) loss will have the best analysis performance. And we performed actual analysis on three data sets that are Subbytes operation part of AES-128 1 round. We verified our prediction by non-profiling analyzing two data sets with a total 16 of MLP-based model, which we describe.

• Journal of the Korea Society of Computer and Information
• /
• v.15 no.3
• /
• pp.1-10
• /
• 2010

As processor platforms are continuously moving toward wireless mobile systems, embedded mobile processors are expected to perform more and more powerful, and therefore the development of an efficient power management algorithm for these battery-operated mobile and handheld systems has become a critical challenge. It is well known that a memory system is a main performance limiter in the processor point of view. Although many DVFS studies have been considered for the efficient utilization of limited battery resources, recent works do not explicitly show the interaction between the processor and the memory. In this research, to properly reflect short/long-term memory access patterns of the embedded workloads in wireless mobile processors, we propose a memory buffer utilization as a new index of DVFS level prediction. The simulation results show that our solution provides 5.86% energy saving compared to the existing DVFS policy in case of memory intensive applications, and it provides 3.60% energy saving on average.

• Journal of Convergence for Information Technology
• /
• v.10 no.5
• /
• pp.1-7
• /
• 2020

This paper proposes a link adaptation method for Underwater Internet of Things (IoT), which reduces power consumption of sensor nodes and improves the throughput of network in underwater IoT network. Adaptive Modulation and Coding (AMC) technique is one of link adaptation methods. AMC uses the strong correlation between Signal Noise Rate (SNR) and Bit Error Rate (BER), but it is difficult to apply in underwater IoT as it is. Therefore, we propose the machine learning based AMC technique for underwater environments. The proposed Modulation Coding and Scheme (MCS) prediction model predicts transmission method to achieve target BER value in underwater channel environment. It is realistically difficult to apply the predicted transmission method in real underwater communication in reality. Thus, this paper uses the high accuracy BER prediction model to measure the performance of MCS prediction model. Consequently, the proposed AMC technique confirmed the applicability of machine learning by increase the probability of communication success.