• Journal of Advanced Marine Engineering and Technology
• /
• v.35 no.5
• /
• pp.667-674
• /
• 2011

Data networks at sea will be overlapped networks with not only traditional carriers such as RF, satellites but also BWA like wireless LAN, WiBro, and WCDMA in near future. In this paper, an overlapped MANET model for data networks at sea, and a routing protocol (OMH-MW) selecting optimal transmission medium for each hop in routes are proposed. OMH-MW measures the optimality of each medium regarding the transmission characteristics of each application and those of the medium in together. The most suitable medium to each link is selected as the link in routes. Performances are compared with those of the MWR (Max-Win based Routing protocol searching optimal routes with only one medium).

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.38 no.3
• /
• pp.228-235
• /
• 2010

This paper deals with an autonomous flight controller design problem for a tilt-rotor aircraft in rotary-wing mode. The inner-loop algorithm is designed using the output-based approximate feedback linearization. The model error originated from the feedback linearization is cancelled within allowable tolerance by using single-hidden-layer neural network. According to Lyapunov direct stability theory, the adaptive update law is derived to run the neural network on-line, which is based on the linear observer dynamics. Moreover, the outer-loop algorithm is designed to track the trajectory generated from way-point guidance. Especially, heading and flight-path angle line-of-sight guidance are applied to the outer-loop to improve accuracy of the landing tracking performance. The 6-DOF nonlinear simulation shows that the overall performance of the flight control algorithm is satisfactory even though the collective input response shows instantaneous actuator saturation for a short time due to the lack of the neural network and the saturation protection logic in that loop.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.12 no.3
• /
• pp.201-207
• /
• 2012

A RAM-based neural network is a weightless neural network based on binary neural network(BNN) which is efficient neural network with a one-shot learning. RAM-based neural network has multiful information bits and store counts of training in BNN. Supervised learning based on the RAM-based neural network has the excellent performance in pattern recognition but in pattern categorization with unsupervised learning as unsuitable. In this paper, we propose a unsupervised learning algorithm in the RAM-based neural network to perform pattern categorization. By the proposed unsupervised learning algorithm, RAM-based neural network create categories depending on the input pattern by itself. Therefore, RAM-based neural network for supervised learning and unsupervised learning should proof of all possible complex models. The training data for experiments provided by the MNIST offline handwritten digits which is consist of 0 to 9 multi-pattern.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2010.10a
• /
• pp.701-704
• /
• 2010

In studying an autonomous navigation technique of UUV(Unmaned Underwater Vehicle), one of the many fundamental techniques is to plan a 3D path to complete the mission via realtime information received by sonar showing landscapes and obstacles. The simulation system is necessary to verify the algorithm in researching and developing 3D path planning of UUV. It is because 3D path planning of UUV should consider guide control, the dynamics, ocean environment, and search sonar models on the basis of obstacle avoidance technique. The simulation system developed in this paper visualizes the UUV's movement of avoiding obstacles, arriving at the goal position via waypoints by using C++ and OpenGL. Plus, it enables the user to setup the various underwater environment and obstacles by a user interface. It also provides a generalization that can verify path planning algorithm of UUV studied in any developing environment.

• Korean Journal of Construction Engineering and Management
• /
• v.12 no.5
• /
• pp.35-45
• /
• 2011

Establishing housing supply strategies in Korean housing market is a crucial issue due to contradictory but concurrent two problems in market; one is the unstable working-class residential and the other is the high vacancy rate by the low-level of sales rate. Although government has been continuously implementing various supply policies in an attempt to evenly distribute houses as well as to keep supply and demand in balance, it is difficult to satisfy all of stakeholders, such as housing consumers, housing owners and housing suppliers. This paper, therefore, applies a system dynamics methodology and offers a dynamic and integrated model encompassing for-profit behaviors of each market participants. The proposed model simulates the future trends of house prices, the balances between supply and demand, construction companies earnings and vacancy rate when applying various housing supply scenarios. From the simulation result, recent governmental small-size rental housing supplies in bulks should utilize private construction companies to stabilize housing distribution rate and private supply system as well as the supply and demand are well balanced.

• Journal of the Korea Convergence Society
• /
• v.9 no.1
• /
• pp.9-20
• /
• 2018

The purpose of this paper is to analyze the effect of corporation internal reservation on R&D investment. It is to find how much effect the reflux tax has achieved through empirical analysis. In addition, education training expense was taken as a moderating variable to find the effectiveness of government policy. Furthermore, the study looked through the effect once again by using GMB model. According to the result counted by regression analysis, it could be concluded that the effect of both moderation and intervention had a significant effect and the variable of interest cost and welfare & benefit cost in model 1, 2 and 3 had a meaningful impact at the level of 99%. On the other hand, the previous corporate internal reservation failed to show any significant result in all types of models. Even in GBM model of convergence level applied to additional analysis, similar results came out.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.12
• /
• pp.10-19
• /
• 1999

Cellular Automata Neural Systems(CANS) are neural networks based on biological development and evolution. Each neuron of CANS has local connection and acts as a form of pulse according to the dynamics of the chaotic neuron. CANS are generated from initial cells according to the CA rule. In the previous study, to obtain the useful ability of CANS, we make the pattern of initial cells evolve. However, it is impossible to represent all solution space, so we propose an evolving method of CA rule to overcome this defect in this paper. DNA coding has the redundancy and overlapping of gene and is apt for the representation of the rule. In this paper, we show the general expression of CA rule and propose translation method from DNA code to CA rule. The effectiveness of the proposed scheme was verified by applying it to the navigation problem of autonomous mobile robot.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2019.05a
• /
• pp.238-241
• /
• 2019

In this paper, a "Vehicle Analysis System(VAS) using cloud and data mining" is proposed that store all the sensor data measured in the vehicle in the cloud, analyze the stored data using the classification model, and provide the analyzed data in real time to the driver's display. The VAS consists of two modules. First, Sensor Data Communication Module(SDCM) stores the sensor data measured in the vehicle in a table of the cloud server and transfers the stored data to the analysis module. Second, Sensor Data Analysis Module(SDAM) analyzes the received data using the genetic algorithm and provides analyzed result to the driver in real time. The VAS stores sensor data collected in the vehicle in the cloud server without accumulating it in the vehicle, and stored data is analyzed in the cloud server, so that the sensor data can be quickly and efficiently managed without overloading the vehicle. In addition, the information desired by the driver can be visualized on the display, thereby increasing the stability of the autonomous vehicle.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2022.11a
• /
• pp.137-139
• /
• 2022

As cyber threats in the maritime industry increase due to the digital transformation, the needs for cyber security training for ship's crew and port engineers has increased. The training of seafarers is related to the IMO's STCW convention, so cyber security training also managed and certified, and it is necessary to develop a cybersecurity training system that reflects the characteristics of the OT systemof ships and ports. In this paper, with the goal of developing a training model based on the IMO cyber risk management guideline, developing a cyber security training model based on the characteristics of maritime industry threats, and improving the effectiveness of cyber security training using AR/VR and metaverse, A method for developing a system for nurturing cyber security experts is presented.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.8
• /
• pp.23-32
• /
• 2024

NVS (Novel View Synthesis) is a field in computer vision that reconstructs new views of a scene from a set of input views. Real-time rendering and high performance are essential for NVS technology to be effectively utilized in various applications. Recently, 3D-GS (3D Gaussian Splatting) has gained popularity due to its faster training and inference times compared to those of NeRF (Neural Radiance Fields)-based methodologies. However, since 3D-GS reconstructs a 3D (Three-Dimensional) scene by splitting and cloning (Density Control) Gaussian points, the number of Gaussian points continuously increases, causing the model to become heavier as training progresses. To address this issue, we propose two methodologies: 1) Gaussian blending, an improved density control methodology that removes unnecessary Gaussian points, and 2) a performance enhancement methodology using a depth estimation model to minimize the loss in representation caused by the blending of Gaussian points. Experiments on the Tanks and Temples Dataset show that the proposed methodologies reduce the number of Gaussian points by up to 4% while maintaining performance.