• Proceedings of the Korea Association of Information Systems Conference
• /
• 2004.11a
• /
• pp.409-416
• /
• 2004

동적 공급사슬망은 복잡하고 다양한 이해관계를 가진 기업들로 구성되어 있다. 다수의 구매자로부터 주문 의뢰가 동시다발적으로 발생하므로 하위 구성원들은 경쟁적 관계에 놓이게 된다. 그러므로 최적의 공급사슬구성을 위해서는 수평적 경쟁 관계를 고려하여 구성주체들간의 협력관계를 통해 이를 해결하여야 한다. 지금까지의 스케줄링 문제에서는 상위의 구성원이 하위 구성원들을 일방적으로 선택하는 의사결정이 이루어졌으나 본 문제에서는 구성원간의 협력관계에서 에이전트를 통한 다자간 협상을 통해 공급사슬 전체의 최적화를 구성하는 방법론을 제시한다. 본 협상방법론은 단일기계에서 상이한 납기일, 조기생산(earliness), 지연생산(tardiness)을 동시에 고려하였으며 전체 공급사슬의 평균절대편차(Mean Absolute Deviation)의 최소화를 목적으로 하고 있다. 본 협상방법론의 효과성을 증명하기 위해 분지한계법(Branch & Bound)과 비교하고, 알고리즘 구현을 통해 구매자 협상방법론의 최적화 여부를 실험을 통해 증명하였다.

• KIPS Transactions on Computer and Communication Systems
• /
• v.6 no.5
• /
• pp.219-230
• /
• 2017

General-Purpose Graphics Processing Units (GPGPUs) build massively parallel architecture and apply multithreading technology to explore parallelism. By using programming models like CUDA, and OpenCL, GPGPUs are becoming the best in exploiting plentiful thread-level parallelism caused by parallel applications. Unfortunately, modern GPGPU cannot efficiently utilize its available hardware resources for numerous general-purpose applications. One of the primary reasons is the inefficiency of existing warp/thread block schedulers in hiding long latency instructions, resulting in lost opportunity to improve the performance. This paper studies the effects of hardware thread scheduling policy on GPGPU performance. We propose a novel warp scheduling policy that can alleviate the drawbacks of the traditional round-robin policy. The proposed warp scheduler first classifies the warps of a thread block into two groups, warps with long latency and warps with short latency and then schedules the warps with long latency before the warps with short latency. Furthermore, to support the proposed warp scheduler, we also propose a supplemental technique that can dynamically reduce the number of streaming multiprocessors to which will be assigned thread blocks when encountering a high contention degree at the memory and interconnection network. Based on our experiments on a 15-streaming multiprocessor GPGPU platform, the proposed warp scheduling policy provides an average IPC improvement of 7.5% over the baseline round-robin warp scheduling policy. This paper also shows that the GPGPU performance can be improved by approximately 8.9% on average when the two proposed techniques are combined.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.9 no.3
• /
• pp.85-93
• /
• 2010

This paper proposes a new dynamic hardware allocation and binding algorithm of a simultaneous allocation and binding for SOC design automation. The proposed algorithm works on scheduled input graph and simultaneously allocates binds functional units, interconnections and registers by considering interdependency between operations and storage elements in each control step, in order to share registers and interconnections connected to functional units, as much as possible. This paper shows the effectiveness of the proposed algorithm by comparing experiments to determine number of function unit in advance or by comparing separated executing allocation and binding of existing system.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.1A
• /
• pp.24-33
• /
• 2007

In this paper, two different dynamic cell coordination strategies for frequency flat and selective fading are proposed for efficient subcarrier allocation in the joint consideration of adaptive modulation and variable frequency reuse in the channel-aware OFDMA downlink multicellular environment. Compared to a conventional OFDMA system without cell coordination, where system throughput may become degraded due to the persistent interference from other cells, the proposed system dynamically allows RNC to apply different reuse factors on each subchannel and scheduling in consideration of channel and interference conditions of individual users so as to increase the system throughput and guarantee QoS of each user. In a frequency flat fading, the dynamic scheme with the proposed scheduling achieves on average three times larger throughput than the conventional dynamic scheme [8]. In a selective fading channel, the proposed schemes showed 2.6 times as large throughput as that of a single reuse factor of one for all subchannels.

• Journal of KIISE:Databases
• /
• v.27 no.2
• /
• pp.227-237
• /
• 2000

A Web database System is a large-scaled multimedia application system that has multimedia processing facilities and cooperates with relational/Object-Oriented DBMS. Conventional hypertext modeling methods and DB gateway have limitations for Web database because of their restricted versatile presentation abilities and inefficient concurrency control caused by bottleneck in cooperation processing. Thus, we suggest a Dynamic Navigation Model & Virtual Graph Structure. The Dynamic Navigation Model supports implicit query processing and dynamic creation of navigation spaces, and introduce node-link creation rule considering navigation styles. We propose a mapping methodology between the suggested hypertext model and the relational data model, and suggest a dynamic allocation scheduling technique for query processing server based on weighted value. We show that the proposed technique enhances the retrieval performance of Web database systems in processing complex queries concurrently.

• The KIPS Transactions:PartC
• /
• v.16C no.4
• /
• pp.477-486
• /
• 2009

This paper proposes a dynamic bandwidth allocation (DBA) algorithm with adaptive maximum transmission window (DBA-AMTW) to increase channel utilization in Ethernet passive optical networks (EPONs). A polling mechanism in EPON determines channel utilization and puts constraints on DBA algorithm and scheduling. DBA algorithms based on interleaved polling with stop (IPS) allocate transmission windows to optical network units (ONU) considering requests of all ONUs. However channel idle time when any ONU does not transmit packets decreases channel utilization. Proposed DBA-AMTW improves efficiency of a network and allocates transmission windows effectively by appropriate DBA computation from REPORT messages of all ONUs. An adaptive maximum transmission window for each ONU determined by a DBA computation in the previous scheduling cycle. Simulation results show that the proposed DBA algorithm improves performance of throughput and average delay time.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.43 no.3 s.345
• /
• pp.135-143
• /
• 2006

PQ(Priority Queue) and WRR(Weighted Round Robin) are the most famous schedulers, however, these schedulers have both points of advantages and disadvantages. In this paper, we propose an algorithm that can be adopted in my kind of scheduling type with making up for weak points of PQ and WRR. The proposed algorithm includes a fuzzy theory in the scheduler to assign priorities dynamically in the DiffServ network. This algorithm produces the control discipline by the fuzzy theory to assign priorities of the Queue effectively with checking the Queue status of each class dynamically under the discipline.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.11A
• /
• pp.859-866
• /
• 2003

In this paper, we designed the bandwidth allocation scheme with priority consideration for upstream channel access of EthernetPON. The objective of our scheme is to control the multi services in more effective way according to their CoS(Class of Service) or QoS(Quality of Service). The designed scheme considers transmission priority in the both side of OLT and ONU. In the OLT's view, the Two-step scheduling algorithm is applied with which we can support multiple bandwidth allocation policies simultaneously, i.e. SBA for the time-sensitive, constant rate transmission services and DBA for the best-effort services. This Two-step scheduling algorithm reduces the scheduling complexity by separating the process of transmission start time decision from the process of grant generation. In the ONU's view, the proposed scheme controls 8 priority queues of the 802.1d recommended 8 service classes. Higher priority queue is serviced in prior during the allowed GATE time from OLT. The OPNET modeling and simulation result compares the performance of each bandwidth allocation policy with SBA or DBA only approach.

• The KIPS Transactions:PartD
• /
• v.11D no.2
• /
• pp.339-350
• /
• 2004

A broadcast method in mobile computing environments which have very large client populations solves the problem of low bandwidth. But most of previous proposed concurrency control protocols are restricted to read-only transactions from mobile client. In this paper, we propose the UTSM-ARM method which uses caching and broadcast method, and allows update transactions in mobile client. The UTSM-ARM decides the validation of cache data consistency with the dynamic update pattern ratio of accessed cached data. Also, the timestamps of accessed cached data and transaction in mobile client guarantee the serializable execution of mobile transactions. As a result, UTSM-ARM makes efficiently use of the asymmetric bandwidth of broadcast environment and can Increase the transaction throughput.

• Journal of KIISE:Software and Applications
• /
• v.32 no.11
• /
• pp.1021-1028
• /
• 2005

The general solution for classification and regression problems can be found by matching and modifying matrices with the information in real world and then these matrices are teaming in neural networks. This paper treats primary space as a real world, and dual space that Primary space matches matrices using kernel. In practical study, there are two kinds of problems, complete system which can get an answer using inverse matrix and ill-posed system or singular system which cannot get an answer directly from inverse of the given matrix. Further more the problems are often given by the latter condition; therefore, it is necessary to find regularization parameter to change ill-posed or singular problems into complete system. This paper compares each performance under both classification and regression problems among GCV, L-Curve, which are well known for getting regularization parameter, and kernel methods. Both GCV and L-Curve have excellent performance to get regularization parameters, and the performances are similar although they show little bit different results from the different condition of problems. However, these methods are two-step solution because both have to calculate the regularization parameters to solve given problems, and then those problems can be applied to other solving methods. Compared with UV and L-Curve, kernel methods are one-step solution which is simultaneously teaming a regularization parameter within the teaming process of pattern weights. This paper also suggests dynamic momentum which is leaning under the limited proportional condition between learning epoch and the performance of given problems to increase performance and precision for regularization. Finally, this paper shows the results that suggested solution can get better or equivalent results compared with GCV and L-Curve through the experiments using Iris data which are used to consider standard data in classification, Gaussian data which are typical data for singular system, and Shaw data which is an one-dimension image restoration problems.