• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.2
• /
• pp.91-96
• /
• 2013

The application field of supercomputing systems are changing to support into the field for both a large-volume data processing and high-performance computing at the same time such as bio-applications. These applications require high-performance distributed file system for storage management and efficient high-speed processing of large amounts of data that occurs. In this paper, we introduce MAHA-FS for supercomputing systems for processing large amounts of data and high-performance computing, providing excellent metadata operation performance and IO performance. It is shown through performance analysis that MAHA-FS provides excellent performance in terms of the metadata processing and random IO processing.

• Journal of Internet Computing and Services
• /
• v.24 no.6
• /
• pp.1-11
• /
• 2023

Federated learning (FL) is a ground breaking machine learning paradigm that allow smultiple participants to collaboratively train models in a cloud environment, all while maintaining the privacy of their raw data. This approach is in valuable in applications involving sensitive or geographically distributed data. However, one of the challenges in FL is dealing with heterogeneous and non-independent and identically distributed (non-IID) data across participants, which can result in suboptimal model performance compared to traditionalmachine learning methods. To tackle this, we introduce FedGCD, a novel FL algorithm that employs Graph Neural Network (GNN)-based community detection to enhance model convergence in federated settings. In our experiments, FedGCD consistently outperformed existing FL algorithms in various scenarios: for instance, in a non-IID environment, it achieved an accuracy of 0.9113, a precision of 0.8798,and an F1-Score of 0.8972. In a semi-IID setting, it demonstrated the highest accuracy at 0.9315 and an impressive F1-Score of 0.9312. We also introduce a new metric, nonIIDness, to quantitatively measure the degree of data heterogeneity. Our results indicate that FedGCD not only addresses the challenges of data heterogeneity and non-IIDness but also sets new benchmarks for FL algorithms. The community detection approach adopted in FedGCD has broader implications, suggesting that it could be adapted for other distributed machine learning scenarios, thereby improving model performance and convergence across a range of applications.

• International Journal of CAD/CAM
• /
• v.3 no.1_2
• /
• pp.51-60
• /
• 2003

The growth of the World Wide Web and the advances in high-speed network access have greatly changed existing CAD/CAE environment. The WWW has enabled us to share various distributed product data and to collaborate in the design process. An international standard for the product model data, STEP, and a standard for the distributed object technology, CORBA, are very important technological components for the interoperability in the advanced design and manufacturing environment. These two technologies provide background for the sharing of product data and the integration of applications on the network. This paper describes a distributed CAD/CAE environment that is integrated on the network by CORBA and product model data standard STEP. Several prototype application modules were implemented to verify the proposed concept and the test result is discussed. Finite element analysis server are further distributed into several frontal servers for the implementation of distributed parallel solution of finite element system equations. Distributed computation of analysis server is also implemented by using CORBA for the generalization of the proposed method.

• Journal of KIISE
• /
• v.44 no.7
• /
• pp.726-732
• /
• 2017

In practical distributed sensing and prediction applications over wireless sensor networks (WSN), environmental sensing activities are highly dynamic because of noisy sensory information from moving source signals. The recent distributed online convex optimization frameworks have been developed as promising approaches for solving approximately stochastic learning problems over network of sensors in a distributed manner. Negligence of mobility consequence in the original distributed saddle point algorithm (DSPA) could strongly affect the convergence rate and stability of learning results. In this paper, we propose an integrated sliding windows mechanism in order to stabilize predictions and achieve better convergence rates in cooperative detection of a moving source signal scenario.

• Journal of the Korea Society for Simulation
• /
• v.18 no.1
• /
• pp.35-40
• /
• 2009

The reliability of processing the distributed application becomes more and more important issues in distributed network as the usage of distributed network increases. Because the distributed network applications are processed in the available resources of Internet. It is important factors that the user can efficiently access the available resources and processed results can be delivered to the user of job request. Therefore in this paper, we suggested the survivability factor for sending data safely whether it can be efficiently accessed between the user and available resource which processes the distributed application. And evaluated the performance of the suggested factor using the data transmission time. Also we suggested the transmission algorithm for safe transmission using the survivability factor.

• 한국연소학회:학술대회논문집
• /
• 2003.05a
• /
• pp.75-83
• /
• 2003

Diode laser absorption system is advantageous of their non-invasive nature, fast response time, high sensitivity and real-time measurement capability. Furthermore, recent advances in room-temperature, near-IR and visible diode laser sources for telecommunication, optical data storage applications are enabling combustion diagnostics system based on diode laser absorption spectroscopy. So, combined with fiber-optics and high sensitive detection strategies, compact and portable sensor system are now appearing for a variety of applications. The objective of this research is to take advantage of distributed feed-back diode laser and develope new gas sensing system. It experimentally found out that the wavelength, power characteristics as a function of injection current and temperature. In addition to direct absorption and wavelength modulation spectroscopy have been demonstrated in these experiments and have a bright prospect to this diode laser system.

• ETRI Journal
• /
• v.27 no.1
• /
• pp.13-21
• /
• 2005

Many multicast applications, such as video-on-demand and video conferencing, desire quality of service (QoS) support from an underlying network. The differentiated services (DiffServ) approach will bring benefits for theses applications. However, difficulties arise while integrating native IP multicasting with DiffServ, such as multicast group states in the core routers and a heterogeneous QoS requirement within the same multicast group. In addition, a missing per-flow reservation in DiffServ and a dynamic join/leave in the group introduce heavier and uncontrollable traffic in a network. In this paper, we propose a distributed and stateless admission control in the edge routers. We also use a mobile agents-based approach for dynamic resource availability checking. In this approach, mobile agents act in a parallel and distributed fashion and cooperate with each other in order to construct the multicast tree satisfying the QoS requirements.

• Proceedings of the Korea Society of Information Technology Applications Conference
• /
• 2007.05a
• /
• pp.84-102
• /
• 2007

Security is crucial for the successful deployment of large distributed systems. Many of these systems provide services to people across different administrative domains. The traditional identity-based access control mechanisms are un scalable and difficult to manage. Unlike the closed systems, open systems provide services to people from different security domains. Healthcare systems need to be highly available in order for the patients to get a timely treatment. The health care information should be available to authorized users both inside the administrative domain and outside the domain, such as pharmacies and insurance companies. In this paper, we first analyzed the necessities and advantages of importing attribute-based trust-management models into open distributed systems. Then we reviewed traditional access-control models and presented the basic elements of the new trust-management model.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.6
• /
• pp.1356-1364
• /
• 1997

This paper presents a modelling and implementation scheme of a multi directory which provides location and supplementary information of multiple applications in distributed environments. An extrended directory schema and extended DIT(Directory Information Tree) model is proposed for the multi directory. For implementation scheme, an implementation example adapted for the DFR(Document Filing and Retrieval) is shown. In addition, a configuration model for interworking between the DFR and the multi directory, an association mechanisms between two application's operations are also described. In a distributed environment, the interworking between the DFR and a multi directory implemented by using QUIPU 8.0 of ISODE(ISO Development Environment) directory system is tested. Based on the result of this paper, an extended model ofthe directory system is proposed for providing search operation of various application' objects.

• Journal of Electrical Engineering and Technology
• /
• v.6 no.4
• /
• pp.573-579
• /
• 2011

To meet the scalability and performance requirements of data analyses, which often involve voluminous data, efficient parallel or concurrent algorithms and frameworks are essential. We present a high-performance Korean morphological analyzer which employs the MapReduce framework on the graphics processing unit (GPU). MapReduce is a programming framework introduced by Google to aid the development of web search applications on a large number of central processing units (CPUs). GPUs are designed as a special-purpose co-processor. Their programming interfaces are typically formulated for graphics applications. Compared to CPUs, GPUs have greater computation power and memory bandwidth; however, GPUs are more difficult to program because of the design of their architectures. The performance of the Korean morphological analyzer using the MapReduce framework on the GPU is evaluated in comparison with the CPU-based model. The proposed Korean Morphological analyzer shows promising scalable performance on distributed computing with the GPU.