• Proceedings of the Korea Society for Simulation Conference
• /
• 1999.04a
• /
• pp.222-226
• /
• 1999

이 기종의 다중 시스템 환경에서의 트랜잭션 처리는 전체 참여 노드의 성능을 최대로 이용하기 위해 효과적인 트랜잭션의 분배가 반드시 필요하다. 이를 위해 트랜잭션 처리에 의해 발생하는 다양한 부하들을 참조해야 한다. 본 논문에서는 효율적인 트랜잭션 분배를 위해 분산 로킹 정보인 주사본 권한(PCA) 정보를 이용한다. 주사본 권한 정보를 이용해 트랜잭션 처리 데이터의 참조 지역성을 극대화하고, 이를 통해 캐쉬 효과를 증가시킨다. 즉 각 노드에 저장된 데이터를 자신의 노드와 나머지모드에 분배해 캐싱하고 이를 이용해 디스크 접근을 최소화 한다. 트랜잭션 분배기는 전역적인 캐싱 정보를 관리하고 이를 이용해 전체 노드들이 평균적인 트랜잭션 처리율을 유지할 수 있도록 트랜잭션을 분배한다. 제안된 기법의 검증을 위해 트랜잭션 처리 시스템을 웹 기반 환경에서 시뮬레이션할 수 있도록 Simjava를 이용해 모델을 구상한다.

• Journal of information and communication convergence engineering
• /
• v.3 no.2
• /
• pp.90-92
• /
• 2005

As the result of the study shows, Round Robin Algorithm allows servers to ensure definite traffic distribution, unless incoming data loads differ much. Although error levels are high in some cases, they were eventually alleviated by repeated tests for a long period of time. The study uses two software packages (Packet Capture and Round Robin Test Package) to check packet volume from Virtual Network Structure (data generator, virtual server, Server 1, 2, 3), and finds out traffic distribution toward Server 1, 2, and 3. The functions of implemented Round Robin Load Balancing Monitoring System include round robin testing, system monitoring, and graphical indication of data transmission and packet volume.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.10a
• /
• pp.526-528
• /
• 2004

메타데이타에 대한 접근이 특정 디렉토리에 집중되며 메타데이타 연산마다 다른 계산량을 가지는 클러스터 파일 시스템의 특성상 메타데이타 서버 간 부하의 불균형과 과부하가 발생한다. 따라서 클러스터 파일 시스템의 성능을 결정짓는 중요 요소인 메타데이타 서비스의 성능을 위해서는 메타데이타 서버들의 과부하 상황에 대처할 수 있는 합리적인 부하 분산 기법이 필수적이다. 메타데이타 공간을 분할하여 담당영역만을 관리하는 비대칭 메타데이타 서버를 위해 본 눈문은 클라이언트 요청의 내용을 분석하여 담당 메타데이타 서버를 결정하고 해당 연산의 종류에 따라 단순 검색, 메타데이타 중복 저장(replication), 또는 메타데이타에 대한 로깅(logging)을 수행하는 내용 기반의 부하 분산 알고리즘을 제시하였다.

• Annual Conference of KIPS
• /
• 2012.11a
• /
• pp.785-788
• /
• 2012

There has been growing interest in multicast technologies for wireless mesh networks with increasing demand for enjoying seamless multimedia broadcast service such as mobile IPTV anytime anywhere for any content on any device. This paper addressed the factors that should be considered when designing multicast algorithm in WMNs. Several routing metrics which are adapted for WMNs based multicast are described in detail to show that the main design goal for multicast metric is to achieve high throughput. Those metrics that take into account the characteristics of WMNs such as wireless broadcast advantage or channel diversity can arrange load balancing more effectively.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.01a
• /
• pp.13-15
• /
• 2023

본 논문에서는 성격, 성향을 근거로 사람의 성향에 따른 직업 및 전공에 대한 만족도를 분류한 데이터셋을 구축하여 사전에 사용자의 성향을 파악하여 직업을 추천하는 알고리즘을 제안한다. 성격유형검사 뿐만이 아닌 최근 게시한 SNS 텍스트를 사전에 학습한 데이터셋에 적용해 성격유형 결과의 정확도를 상승시키고자 한다. 사전에 생성한 데이터셋 외에 대상자가 작성한 정보(직업, 전공, 직엄 및 전공에 대한 만족도)로 연합학습을 진행하여 데이터셋의 정확도를 향상시키고자 한다. 모델의 학습 및 분류의 정확도 향상을 위해 SVM, NB, KNN, SDG 알고리즘들을 비교하였고 각각 67%, 21%, 28%, 69%의 정확도를 도출하였다. 데이터 셋은 캐글에서 제공받았다.

• International Journal of Internet, Broadcasting and Communication
• /
• v.16 no.2
• /
• pp.161-168
• /
• 2024

As the size of big data models grows, distributed training is emerging as an essential element for large-scale machine learning tasks. In this paper, we propose ParamHub for distributed data training. During the training process, this agent utilizes the provided data to adjust various conditions of the model's parameters, such as the model structure, learning algorithm, hyperparameters, and bias, aiming to minimize the error between the model's predictions and the actual values. Furthermore, it operates autonomously, collecting and updating data in a distributed environment, thereby reducing the burden of load balancing that occurs in a centralized system. And Through communication between agents, resource management and learning processes can be coordinated, enabling efficient management of distributed data and resources. This approach enhances the scalability and stability of distributed machine learning systems while providing flexibility to be applied in various learning environments.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.12A
• /
• pp.1160-1170
• /
• 2010

Existing multipath routing mechanism has much overhead to maintain the state of nodes on the multipath route and does not guarantees completely disjoint multipath construction from source to destination. In this paper, we propose an Explicit Disjoint Multipath (EDM) routing algorithm to enhance energy efficiency through removing the flooding mechanism for route discovery process, minimizing the number of nodes participating in route update and balancing the traffic load for entire network. EDM constructs logical pipelines which can create disjoint multipaths in logical way. Then it physically performs anchor node based geographic routing along the logical pipeline in order to build multipath to the destination. EDM can provide the distribution effect of traffic load over the network, help to balance the energy consumption and therefore extend the network lifetime.

• Journal of the Korea Society of Computer and Information
• /
• v.12 no.5
• /
• pp.131-137
• /
• 2007

Recently, Ant Colony Optimization (ACO) is emerged as a simple yet powerful optimization algorithm for routing and load-balancing of both wired and wireless networks. However, there are few researches trying to adopt ACO to enhance routing performance in WSN owing to difficulties in applying ACO to WSN because of stagnation effect. In this paper, we propose an energy-efficient path selection algorithm based on ACO for WSN. The algorithm is not by simply applying ACO to routing algorithm but by introducing a mechanism to alleviate the influence of stagnation. By the simulation result, the proposed algorithm shows better performance in data propagation delay and energy efficiency over Directed Diffusion which is one of the outstanding schemes in multi-hop flat routing protocols for WSN. Moreover, we checked that the proposed algorithm is able to mitigate stagnation effect than simple ACO adoption to WSN.

• Journal of KIISE:Computer Systems and Theory
• /
• v.29 no.8
• /
• pp.445-455
• /
• 2002

In this paper, we present a new task scheduling scheme ior the efficient execution of the tasks of compound nodes of hierarchical task graph(HTG) on shared memory system. The proposed scheme for exploitation functional parallelism is autoscheduling that performs the role of scheduling by processor itself without any dedicated global scheduler. To adapt the proposed scheduling scheme for various platforms, Including a uni-processor systems, Java threads were used for implementation, and the performance is analyzed in comparison with a conventional bit vector method. The experimental results showed that the proposed method was found to be more efficient in its execution time and exhibited good load-balancing when using the experimental parameter values. Furthermore, the memory size could be reduced when using the proposed algorithm compared with a conventional scheme.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.22 no.1
• /
• pp.132-141
• /
• 1998

A parallel version of sheet forming analysis program was developed. This version is compatible with any parallel computers which support MPI that is one of the most recent and popular message passing libraries. For this purpose, SERI-SFA, a vector version which runs on Cray Y-MP C90, a sequential vector computer, was used as a source code. For the sake of the effectiveness of the work, the parallelization was focused on the selected part after checking the rank of CPU consumed from the exemplary calculation on Cray Y-MP C90. The subroutines associated with contact algorithm was selected as targe parts. For this work, MPI was used as a message passing library. For the performance verification, an oil pan and an S-rail forming simulation were carried out. The performance check was carried out by the kernel and total CPU time along with theoretical performance using Amdahl's Law. The results showed some performance improvement within the limit of the selective paralellization.