• The KIPS Transactions:PartA
• /
• v.14A no.7
• /
• pp.451-456
• /
• 2007

In this paper, we regard computer systems as heterogeneous multi-server systems and propose a cumulative fair scheduling scheme that pursues long-term fairness. GPS(generalized processor sharing)-based scheduling algorithms, which are usually employed in single-server systems, distribute available capacity in an instantaneous manner. However, applying them to heterogeneous multi-server systems may cause unfairness, since they may not prevent the accumulation of scheduling delays and the extra capacities are distributed in an instantaneous manner. In our scheme, long-term fairness is pursued by proper distribution of extra capacities while guaranteeing reserved capacities. A reference capacity model to determine the ideal progresses of applications is derived from long-term observations, and the scheduler makes the applications gradually follow the ideal progresses while guaranteeing their reserved capacities. A heuristic scheduling algorithm is proposed and the scheme is examined by simulation.

• Journal of KIISE:Software and Applications
• /
• v.31 no.9
• /
• pp.1246-1257
• /
• 2004

To find an optimal solution with genetic algorithm, it is desirable to maintain the population sire as large as possible. In some cases, however, the cost to evaluate each individual is relatively high and it is difficult to maintain large population. To solve this problem we propose a novel genetic algorithm based on fuzzy clustering, which considerably reduces evaluation number without any significant loss of its performance by evaluating only one representative for each cluster. The fitness values of other individuals are estimated from the representative fitness values indirectly. We have used fuzzy c-means algorithm and distributed the fitness using membership matrix, since it is hard to distribute precise fitness values by hard clustering method to individuals which belong to multiple groups. Nine benchmark functions have been investigated and the results are compared to six hard clustering algorithms with Euclidean distance and Pearson correlation coefficients as fitness distribution method.

• Asian-Australasian Journal of Animal Sciences
• /
• v.6 no.4
• /
• pp.497-507
• /
• 1993

Four truncation selection schemes (SSs) were framed to predict and compare the age structures and genetic responses under the influence of various factor employing the scheme-specific algorithms. Two paths of selection, sires (bulls' sires) and dams (bulls' dams) to breed young bulls were considered. Among variable factors, four levels (0.3, 0.5, 0.7, 0.9) of precision of evaluation, five levels (0.0, 0.05, 0.10, 0.15, 0.20 genetic standard deviation) of genetic differences among age classes and 4 levels of proportions selected (for bulls' sire, 0.05, 0.10, 0.125, 0.25, and for bulls' dams 0.02, 0.04, 0.05, and 0.10) contemplated on both paths of selection. The number of age classes for bulls' dams and bulls' sires were 4 or 8 and 2 or 4, respectively. The stayability across age classes for bulls' dams was assumed to be 0.80 or 0.60. The candidates for selection for bulls' sires were equally distributed (0.5 or 0.25) across the age classes. The SS1 (selection on same proportions as candidates' distribution) revealed longest generation lengths and lowest yearly genetic responses. The average ages were youngest and yearly genetic responses were highest in SS4 (selection at each age-specific truncation point with the same average genetic superiority of selected parents across the ages) and followed by SS3 (selection at each agespecific truncation point with same predicted genetic values) and SS2 (selection at common truncation point on phenotypic values) in a population with overlapping generations. The results revealed the importance of choosing suitable selection scheme to acquire maximum yearly genetic responses especially when the genetic differences among age classes are large and the precision of evaluation is relatively low.

• Journal of Computing Science and Engineering
• /
• v.2 no.3
• /
• pp.249-273
• /
• 2008

Epidemic protocols have two fundamental assumptions. One is the availability of a mechanism that provides each node with a set of log(N) (fanout) nodes to gossip with at each cycle. The other is that the network size N is known to all member nodes. While it may be trivial to support these assumptions in small systems, it is a challenge to realize them in large open dynamic systems, such as peer-to-peer (P2P) systems. Technically, since the most fundamental parameter of epidemic protocols is log(N), without knowing the system size, the protocols will be limited. Further, since the network churn, frequently observed in P2P systems, causes rapid membership changes, providing a different set of log(N) at each cycle is a difficult problem. In order to support the assumptions, the fanout nodes should be selected randomly and uniformly from the entire membership. This paper investigates one possible solution which addresses both problems; providing at each cycle a different set of log(N) nodes selected randomly and uniformly from the entire network under churn, and estimating the dynamic network size in the number of nodes. This solution improves the previously developed distributed algorithm called Shuffle to deal with churn, and utilizes the Shuffle infrastructure to estimate the dynamic network size. The effectiveness of the proposed solution is evaluated by simulation. According to the simulation results, the proposed algorithms successfully handle network churn in providing random log(N0 fanout nodes, and practically and accurately estimate the network size. Overall, this work provides insights in designing epidemic protocols for large scale open dynamic systems, where the protocols behave autonomically.

• Journal of KIISE:Computer Systems and Theory
• /
• v.36 no.3
• /
• pp.149-157
• /
• 2009

The crossed cube, a variation of the hypercube, possesses a better topological property than the hypercube in its diameter that is about half of that of the hypercube. It has been known that an N-node complete binary tree is a subgraph of an (N+1)-node crossed cube [P. Kulasinghe and S. Bettayeb, 1995]. However, efficient embedding methods have not been known for the case that the number of nodes of the complete binary tree is greater than that of the crossed cube. In this paper, we show that an N-node complete binary tree can be embedded into an M-node crossed cube with dilation 1 and load factor [N/M], N>M$\geq$2. The dilation and load factor is optimal. Our embedding has a property that the tree nodes on the same level are evenly distributed over the crossed cube nodes. The property is especially useful when tree-structured algorithms are processed on a crossed cube in a level-by-level way.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.5
• /
• pp.1063-1070
• /
• 2015

In this paper, we propose a distributed user scheduling with transmit power control based on the amount of generating interference to other base stations (BSs) in multi-cell multi-input multi-output (MIMO) networks. Assuming that the time-division duplexing (TDD) system is used, the interference channel from users to other cell BSs is obtained at each user. In the proposed scheduling, each user first generates a transmit beamforming vector by using singular value decompositon (SVD) over MIMO channels and reduces the transmit power if its generating interference to other BSs is larger than a predetermined threshold. Each BS selects the user with the largest effective channel gains among users, which reflects the adjusted power of users. Simulation results show that the proposed technique significantly outperforms the existing user scheduling algorithms.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.5B
• /
• pp.482-490
• /
• 2004

Recently, XML has been emerging as a standard for storing and exchanging of data for various distributed applications based on the Internet. Since there are increasing demands to store and manage XML documents, a lot of research works are going on this area to develop new took and techniques based on the XHL. However, most of the researches are concentrated on mapping techniques based on instance or DTD, and the main focus is on structural transformation. Current trend of research is toward the usage of XML documents based on XML schema, and demands not only conversion of structure but also preservation of the semantic constraints. This paper sets up the using of DIDL standing on the basis of XML schema from MPEG-21 as an application domain, and proposes the mapping model that can preserve semantic constraints in addition. We expand previous research techniques in the preprocessing step for the specific domain, and then, apply various new mapping methods in the postprocessing step. We present and discuss the system architecture for implementation, and introduce the algorithms and present implementation environment and semantic extension methodology in detail. Finally we show actual table and query processing based on our proposal.

• Journal of Communications and Networks
• /
• v.7 no.3
• /
• pp.325-336
• /
• 2005

Mobile ad hoc networks (MANET) are composed of moving wireless hosts which, within range of each other, form wireless networks. For communication to occur between hosts that are not within each other's range, routes involving intermediate nodes must be established; however, since the hosts may be in motion, a host that was part of a route may move away from its upstream and downstream partners, thus breaking the route. In this paper, we propose anticipated route maintenance (ARM) protocol with two extensions to route discovery based routing scheme: Extend the route when nodes on a link move apart from each other and they have common neighbor that can be 'inserted' in the path, and shrink route when a node discovers that one of its neighbor which is not the next hop is also on the same route several hops later on. By utilizing only local geographic information (now a part of some route finding algorithms), a host can anticipate its neighbor's departure and, if other hosts are available, choose a host to bridge the gap, keeping the path connected. We present a distributed algorithm that anticipates route failure and performs preventative route maintenance using location information to increase a route lifespan. The benefits are that this reduces the need to find new routes (which is very expensive) and prevents interruptions in service. As the density of nodes increases, the chance to successfully utilize our route maintenance approach increases, and so does the savings. We have compared the performance of two protocols, pure dynamic source routing (DSR) protocol and DSR with ARM. The simulation results show how ARM improves the functionality of DSR by preventing the links in the route from breaking. Packets delivery ratio could be increased using ARM and achieved approximately $\100%$ improvement. The simulations clarify also how ARM shows a noticeable improvement in dropped packets and links stability over DSR, even though there is more traffic and channel overhead in ARM.

• The Journal of the Korea Contents Association
• /
• v.14 no.10
• /
• pp.840-849
• /
• 2014

The paper shows that algorithm education positively affects algorithmic thinking for elementary students. Experiments are conducted on thirty five sixth graders without prior experience of algorithm. The experimental group is given eleven sessions of lesson for four weeks in which algorithmic solution is sought for graph coloring problems. After four weeks of experiments questionnaires are distributed to the students in order to measure differences in algorithmic interest and algorithmic thinking ability. The following results are obtained: Firstly, it may be observed that algorithm education with graph coloring may be effective in reinforcing students' interest in algorithms. Secondly, the experiment shows that algorithmic thinking ability may be enhanced from participating in graph coloring activities. In conclusion, algorithm education with graph coloring problem helps students develop algorithmic thinking ability as well as cultivate students' interest in algorithmic thinking.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.48 no.2
• /
• pp.1-8
• /
• 2011

An underwater robot S/W framework consists of various sub-modules such as sensory data processing module, thruster control module, cognition module and behavior control module. Performance of a robot is determined by not only the efficiency of algorithms used but also effectiveness of their implementations. One most important factor of the effective implementation is the efficiency of data sharing module, as it transmits signals and data between the sub-modules and thus is directly related to the cycles of sensing and control The ideal data sharing module enables immediate access to any data source irrespective of system configurations. In reality, however, there are lots of obstacles including limitation of processing capacity of source modules, delay over network, and scheduling latency of operating systems. The paper proposes a new data sharing architecture and programming models to effectively handle such obstacles in implementation of underwater S/W framework on a small scale distributed computing system.