• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.13 no.4
• /
• pp.296-308
• /
• 2001

A fully nonlinear Boussinesq equation of Wei et al. is finite differenced by Adams predictor-corrector method. A spatially distributed source function and sponge layers are used to reduce the reflected waves in the domain and wale breaking mechanism is included in the equation. The generated waves are found to be good and the corresponding wale heights are very close to the target values. The shoaling of solitary wave and transformation of regular wave over submerged shelf were simulated successfully. The characteristics of breaking mechanism was identified through the numerical experiment and the results of two dimensional wave propagation test over the spherical shoal showed the importance of nonlinear wave model.

• Proceedings of the SAREK Conference
• /
• 2005.11a
• /
• pp.439-444
• /
• 2005

Steady state simulation and dynamic simulation were performed to analyze the operational characteristics of a multi-type refrigeration system, Fully distributed model was adopted to simulate the steady state and transient responses of the system. The main aim was to see the effect of one indoor unit on the other unit. Numerical simulations were carried out for various operation conditions of an indoor unit - secondary fluid inlet temperature, mass flow rate and expansion valve opening. The results showed that the inlet temperature and mass flow rate of the secondary fluid of one indoor unit had minor effect on the operation of the other unit. However, the opening of the expansion valve had significant effect on the performance of the other unit.

• Journal of the Korean GEO-environmental Society
• /
• v.16 no.6
• /
• pp.39-43
• /
• 2015

This study extends the Distributed Temperature Sensing (DTS) application to delineate the saturated zones in shallow sediment and evaluate the groundwater flow in both downward and upward directions. Dry, partially and fully saturated zones and water level in the subsurface can be recognized from this study. High resolution seepage velocity in vertical direction was estimated from the temperature data in the fully saturated zone. By a single profile, water level can be detected and seepage velocity in saturated zone can be estimated. Furthermore, thermal gradient analysis serves as a new technique to verify unsaturated and saturated zones in the subsurface. The vertical seepage velocity distribution in the recognized saturated zone is then analyzed with improvement of Bredehoeft and Papaopulos' model. This new approach provides promising potential in real-time monitoring of groundwater movement.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.8
• /
• pp.2691-2707
• /
• 2014

Medium access control is critical in wireless networks for efficient spectrum utilization. In this paper, we introduce a novel collision resolution method based on the technique of known interference cancellation, and propose a new MAC protocol named as CR-MAC, in which AP tries to decode all the collided data packets by combining partial retransmissions and known interference cancellation. As the collided transmissions are fully utilized, less retransmission is required, especially in a crowded network. The NS-2simulation and MATLAB numerical results show that, under various network settings, CR-MAC performs much better than the IEEE 802.11 DCF in terms of the aggregation throughput and the expected packet delay.

• Journal of Communications and Networks
• /
• v.5 no.3
• /
• pp.197-201
• /
• 2003

We study in this paper both distributed power control and removal algorithms for multimedia CDMA wireless networks. In our study, users can have different data rates as well as different quality of service (QoS) requirements. We derive a necessary and sufficient condition for the fully distributed power control (FDPC) algorithm to find a feasible power set. We also prove that, if the maximal power level is used at the start, then the distributed constrained power control (DCPC) algorithm is equivalent to the FDPC algorithm. For the connection removal algorithm, we prove that the non-reinitialized removal algorithm finds a feasible power set faster and employs smaller power levels than the reinitialized one does. Performances of some connection removal criteria are also studied. Our simulation results reveal that the smallest normalized CIR (SNC) and largest CIR requirement (LCR) criteria result in smaller outage probability than the smallest CIR (SC) criterion in a multimedia environment.

• Journal of Korean Port Research
• /
• v.13 no.2
• /
• pp.289-302
• /
• 1999

As the size and complexity of network increase Distributed System Management (DSM) will be significant issue within information network in order to increase the high reliability and to improve the flexibility of network management. The OSI management model has several problems. The key problems are that it does not fully address the problem of how to develop communication protocol in support of DSM and how to classify the management connection criteria. In this paper to solve first problem described above this paper propose the connectionless CMIP to accomplish for effectively managing the distributed management managing systems to handle the dynamic informations. To work out second problem we introduce the connection criteria in the hierarchy of management system and finally evaluate the efficiency of a suggested protocol during cooperative negotiation among the managing system.

• KIPS Transactions on Computer and Communication Systems
• /
• v.10 no.7
• /
• pp.191-198
• /
• 2021

Since the size of training dataset become large and the model is getting deeper to achieve high accuracy in deep learning, the deep neural network training requires a lot of computation and it takes too much time with a single node. Therefore, distributed deep learning is proposed to reduce the training time by distributing computation across multiple nodes. In this study, we propose hybrid allreduce strategy that considers the characteristics of each layer and communication and computational overlapping technique for synchronization of distributed deep learning. Since the convolution layer has fewer parameters than the fully-connected layer as well as it is located at the upper, only short overlapping time is allowed. Thus, butterfly allreduce is used to synchronize the convolution layer. On the other hand, fully-connecter layer is synchronized using ring all-reduce. The empirical experiment results on PyTorch with our proposed scheme shows that the proposed method reduced the training time by up to 33% compared to the baseline PyTorch.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.12 no.19
• /
• pp.89-98
• /
• 1989

This research addresses the most sophisticated level of complexity of current electronic data processing(EDP) technology realtime, distributed data base(DDB) system. The major objective is to develop guidelines for the control and audit of these sophisticated system, since the control and audit philosophies, control features, and audit techniques applicable to traditional EDP systems are no longer fully appropriate for, or relevant to, these stat-of-art systems. An attempt is made to develop an internally consistent audit and control Conceptual Framework summarizing and integrating the material developed in this research, which then leads into the analysis of the control and audit model.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.39 no.2
• /
• pp.1-9
• /
• 2002

The use of Object-Oriented Distributed Programming(OODP) environment such as DCOM, DSOM, Java RMI, CORBA to implement distributed applications is becoming increasingly popular. However, absence of a fault-tolerance feature in these middleware platforms complicates the design and implementation of reliable distributed object-based applications, although they greatly enhance the quality and reusability of the distributed object-based applications. In this paper, we propose a fault-tolerant programming environment based on RMI, namely Evergreen, for the reliable distributed computing with checkpoints and rollback-recovery mechanism. Based on a series of experiments, we evaluate the performance of Evergreen and find its possibility of extension to fully support our optimal design goal.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.2030-2035
• /
• 2005

In this paper, we study the distributed coordination of a set of synchronous, anonymous, memoryless mobile robots that can freely move on a two-dimensional plane but are unable to communicate directly. Based on this model, we analyze the application problem that consists in having a group of robots form a barricade line to protect from car traffic a crowd of demonstrators parading on the street. For the sake of robustness, we privilege fully decentralized solutions to the problem. In particular, we give a self-stabilizing distributed algorithm to address the problem, in this presentation