• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.1
• /
• pp.117-135
• /
• 2016

Performance enhancement is one of the most important issues in high performance distributed computing systems. In such computing systems, online users submit their jobs anytime and anywhere to a set of dynamic resources. Jobs arrival and processes execution times are stochastic. The performance of a distributed computing system can be improved by using an effective load balancing strategy to redistribute the user tasks among computing resources for efficient utilization. This paper presents a multi-class load balancing strategy that balances different classes of user tasks on multiple heterogeneous computing nodes to minimize the per-class mean response time. For a wide range of system parameters, the performance of the proposed multi-class load balancing strategy is compared with that of the random distribution load balancing, and uniform distribution load balancing strategies using simulation. The results show that, the proposed strategy outperforms the other two studied strategies in terms of average task response time, and average computing nodes utilization.

• Proceedings of the KIEE Conference
• /
• 2001.05a
• /
• pp.42-45
• /
• 2001

The islanding protection for distributed resources (DR) becomes an important and emerging issue in power system protection since the distributed generator installations are rapidly increasing and most of the installed systems are interconnected with distribution network. In order to avoid the negative impacts on distributed network resulting from islanding operations of DR, it is necessary to detect the loss of mains (LOM) effectively. This paper presents a new LOM detection method using the rate of change in total harmonic distortion (THD) of current. The proposed method effectively detects LOM of the DR unit operating in parallel with the distribution network. We also verified the efficiency of the proposed method using the radial distribution network of IEEE 34 bus model.

• International journal of advanced smart convergence
• /
• v.13 no.3
• /
• pp.143-149
• /
• 2024

As the size of data and models in machine learning training continues to grow, training on a single server is becoming increasingly challenging. Consequently, the importance of distributed machine learning, which distributes computational loads across multiple machines, is becoming more prominent. However, several unresolved issues remain regarding the performance enhancement of distributed machine learning, including communication overhead, inter-node synchronization challenges, data imbalance and bias, as well as resource management and scheduling. In this paper, we propose ParamHub, which utilizes orchestration to accelerate training speed. This system monitors the performance of each node after the first iteration and reallocates resources to slow nodes, thereby speeding up the training process. This approach ensures that resources are appropriately allocated to nodes in need, maximizing the overall efficiency of resource utilization and enabling all nodes to perform tasks uniformly, resulting in a faster training speed overall. Furthermore, this method enhances the system's scalability and flexibility, allowing for effective application in clusters of various sizes.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2005.09a
• /
• pp.55-56
• /
• 2005

• Journal of Electrical Engineering and Technology
• /
• v.7 no.6
• /
• pp.834-844
• /
• 2012

Nonlinear electronic loads draw harmonic currents from the power grids that can cause energy loss, miss-operation of power equipment, and other serious problems in the power grids. This paper proposes a harmonic compensation method using multiple distributed resources (DRs) such as small distributed generators (DGs) and battery energy storage systems (BESSs) that are integrated to the power grids through power inverters. For harmonic compensation, DRs should inject additional apparent power to the grids so that certain DRs, especially operated in proximity to their rated power, may possibly reach their maximum current limits. Therefore, intelligent coordination methods of multiple DRs are required for efficient harmonic current compensation considering the power margins of DRs, energy cost, and the battery state-of-charge. The proposed method is based on fuzzy multi-objective optimization so that DRs can cooperate with other DRs to eliminate harmonic currents with optimizing mutually conflicting multi-objectives.

• Korean Journal of Hydrosciences
• /
• v.5
• /
• pp.1-16
• /
• 1994

In this paper distributed models for simulating spatially and temporally varied moving storm in a watershed were developed. The complete simulation in a watershed is achieved through two sequential flow simulations which are overland flow simulation and channel network flow simulation. Two dimensional continuity equation and momentum equation of kinematic approximation were used in the overland flow simulation. On the other hand, in the channel network simulation two types of governing equations which are one dimensional continuity and momentum equations between two adjacent sections in a channel, and continuity and energy equations at a channel junction were applied. The finite difference formulations were used in the channel network model. Macks Creek Experimental Watershed in Idaho, USA was selected as a target watershed and the moving storm on August 23, 1965, which continued from 3:30 P.M. to 5:30 P.M., was utilized. The rainfall intensity fo the moving storm in the watershed was temporally varied and the storm was continuously moved from one place to the other place in a watershed. Furthermore, runoff parameters, which are soil types, vegetation coverages, overland plane slopes, channel bed slopes and so on, are spatially varied. The good agreement between the hydrograph simulated using distributed models and the hydrograph observed by ARS are Shown. Also, the conservations of mass between upstreams and downstreams at channel junctions are well indicated and the wpatial and temporal vaiability in a watershed is well simulated using suggested distributed models.

• Journal of the Korea Society of Computer and Information
• /
• v.11 no.4 s.42
• /
• pp.69-75
• /
• 2006

Internet-based distributed computing environment have been developed for advanced science and engineering by sharing large-scale resources. Therefore efficient scheduling algorithms for allocating user job to resources in the Internet-based distributed computing environment are required. Many scheduling algorithms have been proposed. but these algorithms are not suitable for the Internet-based Distributed computing environment. That is the previous scheduling algorithm does not consider peer self-control. In this paper, we propose a Peer Availability Period Prediction Strategy for Internet-based distributed computing environment and show that our Strategy has better performance than other Strategy through extensive simulation.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2002.05a
• /
• pp.221-226
• /
• 2002

• Proceedings of the Korea Water Resources Association Conference
• /
• 2001.05a
• /
• pp.230-235
• /
• 2001

• Proceedings of the Korea Water Resources Association Conference
• /
• 2005.09a
• /
• pp.793-794
• /
• 2005