• Journal of Information Processing Systems
• /
• v.16 no.4
• /
• pp.832-844
• /
• 2020

The degree of dispersion of a random variable can be described by the variance, which reflects the distance of the random variable from its mean. However, the time complexity of the traditional variance calculation algorithm is O(n), which results from full calculation of all samples. When the number of samples increases or on the occasion of high speed signal processing, algorithms with O(n) time complexity will cost huge amount of time and that may results in performance degradation of the whole system. A novel multi-step recursive algorithm for variance calculation of the time-varying data series with O(1) time complexity (constant time) is proposed in this paper. Numerical simulation and experiments of the algorithm is presented and the results demonstrate that the proposed multi-step recursive algorithm can effectively decrease computing time and hence significantly improve the variance calculation efficiency for time-varying data, which demonstrates the potential value for time-consumption data analysis or high speed signal processing.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.11 no.11
• /
• pp.2185-2191
• /
• 2007

The traditional sensor network is composed of the cable and the sensor of high price, when collecting a sensing data, there is a weak point which is not pliability. WSN uses the equipment of low price, it will be able to collect the data which is diverse from various node. In this paper we composed coal mining topology which used IEEE802.15.4 MAC in Korea Coal Corporation site. We proposed models for the mobility support of the work manager from the coal mining, we selected the optimum model through simulation experiments. When applying the WSN in the Korea Coal Corporation and other mines, this result can be used as a basis.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.11 no.3
• /
• pp.523-531
• /
• 2007

In this paper, we are going to develop a newly designed indexing scheme which is compatible to manage the moving objects and propose a cost model of the scheme. We propose a dynamic hashing index that insertion/delete costs are low. The dynamic hashing structure is that apply dynamic hashing techniques to combine a hash and a tree to a spatial index. We analyzed the dynamic index structure and the cost model by the frequent position update of moving objects and verified through a performance assessment experiment. The results of our extensive experiments show that the newly proposed indexing schemes(Dynamic Hashing Index) are much more efficient than the traditional the fixed grid and R-tree.

• Korean Journal of Remote Sensing
• /
• v.22 no.5
• /
• pp.357-366
• /
• 2006

A traditional image analysis or classification method using satellite imagery is mostly based on the spectral information. However, the spatial information is more important according as the resolution is higher and spatial patterns are more complex. In this study, we attempted to compare and analyze the variogram properties of actual high resolution imageries mainly in the urban area. Through the several experiments, we have understood that the variogram is various according to a sensor type, spatial resolution, a location, a feature type, time, season and so on and shows the information related to a feature size. With simple modeling, we confirmed that the unique variogram types were shown unlike the classical variogram in case of small subsets. Based on the grasped variogram characteristics, we made a level index map for determining urban complexity or land-use classification. These results will become more and more important and be widely applied to the various fields of high-resolution imagery such as KOMPSAT-2 and KOMPSAT-3 which is scheduled to be launched.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2008.11a
• /
• pp.397-400
• /
• 2008

Traditional propellants which have a hypergolic characteristic have a high performance but also have disadvantages of toxicity and complex handling requirement. In order to replace these propellants, one of the alternatives is hydrogen peroxide which generates high temperature oxygen and water vapor after catalytic reaction. In this paper, autoignition characteristics of kerosene by decomposed hydrogen peroxide were investigated to perform fundamental research for designing a thruster using hydrogen peroxide and kerosene propellants. Contraction ratio, whether flame holder exists or not, and feeding pressure of propellants were selected as variables. From the experiments for different mixture ratio, we confirmed the ignition stability is strongly affected by a feeding pressure of propellants.

• Proceedings of the SAREK Conference
• /
• 2006.06a
• /
• pp.860-865
• /
• 2006

The traditional Korean Ondol System that is a radiant floor heating system was made as warm floor and cool indoor temperature. Nowaday, Ondol is developed as the hydronic floor heating system. But unbalance of floor temperature and indoor temperature is occurred bocause strengthen thermal insulation and airtightness in building changes thermal performance. To solve these problems, we examine actual indoor environment of heating system methods in existing apartments and present the new method of floor heating system. The existing heating system made definite indoor temperatures but floor temperatures that is $22^{\circ}C-26^{\circ}C$ was maintained. To solve these problems, we adopted the differential heating system which made warm area and cool area. A differential heating system was made different pitches of heating pipe in single zone and ratio of warm area to cool area is 1 to 2. As a result of experiments, warm area temperature is $40.7^{\circ}C$, cool area temperature is $36.1^{\circ}C$. A difference of temperature between both area is 4K. A distribution of indoor vertical temperature is similar to both warm area and cool area.

• Nuclear Engineering and Technology
• /
• v.52 no.5
• /
• pp.956-963
• /
• 2020

Dose optimization for Radioactive Occupational Personal (ROP) is an important subject in nuclear and radiation safety field. The geometric environment of a nuclear facility is complex and the work area is radioactive, so traditional navigation model and radioactive data field cannot form an effective environment model for dose assessment and dose optimization. The environment model directly affects dose assessment and indirectly affects dose optimization, this is an urgent problem needed to be solved. Therefore, this paper focuses on an environment model used for Dose Assessment and Dose Optimization (DA&DO). We designed a multi-layer radiation field coupling modeling method, and then explored the influence of the environment model to DA&DO by virtual simulation. Then, a simulation test is done, the multi-layer radiation field coupling model for nuclear facilities is demonstrated to be effective for dose assessment and dose optimization through the experiments and analysis.

• Journal of Information Processing Systems
• /
• v.17 no.4
• /
• pp.787-800
• /
• 2021

Three-dimensional (3D) object classification tasks using point clouds are widely used in 3D modeling, face recognition, and robotic missions. However, processing raw point clouds directly is problematic for a traditional convolutional network due to the irregular data format of point clouds. This paper proposes a pointwise convolution neural network (CNN) structure that can process point cloud data directly without preprocessing. First, a 2D convolutional layer is introduced to percept coordinate information of each point. Then, multiple 2D convolutional layers and a global max pooling layer are applied to extract global features. Finally, based on the extracted features, fully connected layers predict the class labels of objects. We evaluated the proposed pointwise CNN structure on the ModelNet10 dataset. The proposed structure obtained higher accuracy compared to the existing methods. Experiments using the ModelNet10 dataset also prove that the difference in the point number of point clouds does not significantly influence on the proposed pointwise CNN structure.

• Transactions of Materials Processing
• /
• v.30 no.6
• /
• pp.275-283
• /
• 2021

A parametric study was conducted on the effects of five fundamental design parameters on springback, including die clearance, step height, step width, punch radius, and taper relief in an L-bending process, controlled by the compression force. The experiment was also conducted to verify the usefulness of the parametric study procedure for process design, as well as the finite element predictions. The elastoplastic finite element method was utilized. The L-bending process of the york product, which is a key part of the breaker mechanism, was employed. The deformation of the material was assumed to be due to plane strain. Five samples of each design parameter were selected based on experiences in terms of process design. The finite element predictions were analyzed in detail to show a shortcut towards the process design improvement which can replace the traditional process design procedure relying on trial-and-errors. The improved process design was verified to meet all the requirements and the predictions and experiments were in good agreement.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.2
• /
• pp.729-748
• /
• 2021

Vocal detection is one of the fundamental steps in musical information retrieval. Typically, the detection process consists of feature extraction and classification steps. Recently, neural networks are shown to outperform traditional classifiers. In this paper, we report our study on how to improve detection accuracy further by carefully choosing the parameters of the deep network model. Through experiments, we conclude that a feature-classifier model is still better than an end-to-end model. The recommended model uses a spectrogram as the input plane and the classifier is an 18-layer convolutional neural network (CNN). With this arrangement, when compared with existing literature, the proposed model improves the accuracy from 91.8% to 94.1% in Jamendo dataset. As the dataset has an accuracy of more than 90%, the improvement of 2.3% is difficult and valuable. If even higher accuracy is required, the ensemble learning may be used. The recommend setting is a majority vote with seven proposed models. Doing so, the accuracy increases by about 1.1% in Jamendo dataset.