In this paper, we propose new algorithms for the partition of input space and the generation of fuzzy control rules. The one consists of Shannon and extended fuzzy entropy function, the other consists of adaptive fuzzy neural system with back propagation teaming rule. The focus of this scheme is to realize the optimal fuzzy rule base with the minimal number of the parameters of the rules, reducing the complexity of the system. The proposed algorithm is tested with the time series prediction problem using Mackey-Glass chaotic time series.

In this paper, we propose the auto-tunning method for the gradient of activation function using fuzzy inference. In order to verify the usefulness of the proposed method, we simulated it with one link manipulator system and confirmed the excellency,

In this paper, we propose hybrid control method of neural network using the 3-point search algorithm. Proposed control method is searched the weight using the 3-point search algorithm for off-line then control the on-line. In order to verify the usefulness of the proposed method, we simulated the proposed control method with one link manipulator system and confirmed the excellency.

This paper presents new on-line handwritten algorithm for continuous alphabet uppercase characters. The algorithm is based on the idea that alphabet uppercase character consists of at most 4 strokes. It tries to determine the maximum output for a recognition result among outputs of four recognizers which have the capacity to discriminate the character using from 1 through 4 stroke information. The recognition module has 4 neural network based recognizers, which can recognize from 1 through 4 stroke character. We also use specialized post-processing techniques for improving the recognition performance. Trained on 440 input data and choosing 390 uppercase words for a recognition test we reached a 92% recognition rate.

In this paper, to perform iris recognition, the iris is changed to 1-D iris signature and methods of efficient iris pattern transformation are discussed. To represent iris signature's frequency characteristics, Fourier transform, Gabor filtering, and wavelet transform are proposed. The consistency between same person's iris and the discrimination between different person's iris are defined by using correlation. Based on these, three transform methods are compared and analyzed.

Many human inventions were inspired by nature. The artificial neural network is one example. Another example is Genetic Algorithms(GA). GAs search by simulating evolution, starting from an initial set of solutions or hypotheses, and generating successive "generations" of solutions. This particular branch of AI was inspired by the way living things evolved into more successful organisms in nature. To simulate the process of GA in a computer, we must simulate many times according to varying many GA parameters. This paper describes the implementation of Java Package for efficient applications on Genetic Algorithms, called "JavaGA". The JavaGA used as a application program as well as applet provides graphical user interface of assigning major GA parameters.

Genetic Algorithm is well known as the efficient algorithm which can solve a difficult optimization problem. Recently, there has been increasing interest in applying genetic algorithm to problem related to network design. In this paper, we propose a two-step genetic algorithm for designing a optimum virtual path network(VPN) for a given physical network and traffic demand. The first step is a routing step in which a route is found between every node pair in the network. In the second step, paths are assigned as VPs so as to minimize the total number of VPs configured, the number of VPs carried by a link, and the VP hopcount. We study the performance of the propose algorithm through simulation. The result shows that the VPN generated by the proposed algorithm is good in minimizing the number of VPs configured, the load on a link, and the VP hopcount.

In this paper, we propose a new method to parallelize multidimensional subscript loop with non-uniform distance. A loop comprise most of the computation in a program and the most important source of parallelism. Multidimensional subscript within a loop difficult to determine for distance to be required loop dependence Elimination. Therefore we Propose new methods that is three algorithm to search subscript.

Extensible Markup Language(XML) is fast emerging as the dominant standard for representing data in the World Wide Web. Sophisticated query engines that allow users to effectively tap the data stored in XML documents will be crucial to exploiting the full power of XML. While there has been a great deal of activity recently proposing new semi-structured data models and query languages for this purpose, this paper explores the more conservative approach of using traditional relational database engines for processing XML documents conforming to Document Type Descriptors(DTDs). In this paper, we describe how to generate relational schemas from XML DTDs. The main issues that must be addressed inc]ode (a) dealing with the complexity of DTD element specifications (b) resolving the conflict between the two-level nature of relational schemas (table and attribute) vs. the arbitrary nesting of XML DTD schemas and (c) dealing with set-valued attributes and recursion. We now propose a set of transformations that can be used to "simplify" any arbitrary DTD without undermining the effectiveness of queries over documents conforming to that DTD.

In this paper, we propose the method to express various factors for software architecture modeling effectively using UML extension mechanism. Furthermore, we show usefulness of proposed method by presenting several modeling techniques and their examples.

In this paper, we propose a new destage algorithms, called the Dynamic Adaptive Threshold which determines turn-on and turn-off thresholds dynamically depending on the current write cache occupancy level and the differential rate of the host write requests. For performance evaluation, the proposed algorithm is compared with the wellknown High-Low Water Mark (HLWM) algorithm. Performance tests are fulfilled with our cached RAID 5 simulator. The simulation results show that the proposed algorithm outperforms the HLWM algorithm in terms of response time of host reads and write cache hit ratio under various workloads.

In this paper, an compiler system is ported and modified for register allocation experiments. This compiler system will enable various global register allocation. Lcc is introduced and Chaitin's graph coloring algorithm is executed with cmcc on DEC ALPHA 255/300. Several functions of SPEC921NT is used as inputs of the compiler system.

To achieve high performance by exploiting instruction level parallelism(ILP) aggressively in superscalar processors, value prediction is used. Value prediction is a technique that breaks data dependences by predicting the outcome of an instruction and executes speculatively it's data dependent instruction based on the predicted outcome. In this paper, the performance of a hybrid value prediction scheme with dynamic classification mechanism is measured and analyzed by using execution-driven simulator for SPECint95 benchmark set.

A software application builder has designed and partially implemented a E-Business software system using several reusable in-house software components. The builder finds an externally available third-party software components that satisfies solve desired functionality or behavior. We need systematic classification of the component from the domain. We propose a reference architecture of E-Business domain. It is used to search and reuse requiring components.

WBI provides new opportunities to realize the flexible learning environment based on hypermedia and to support distance learning with a diverse interaction. The instructors or learners in WBI claim to be able to resolve reluctant fluctuations such as disorientation and cognitive overload. To overcome these phenomena, a supplementary tool able to manage learning space organized by the instructor's or learner's own way and offer effective navigation techniques is presented in this paper. A learning space management and navigation tool called HyperMap dynamically represents the learning space in the form of a two-dimensional labeled graph. This HyperMap also can be used for an instruction design tool, learner's portfolio for the exchange of learning experiences. and the assessment of WBI.

In a distributed real-time system(DRTS), testing and debugging are difficult and critical procedures since they implies several problems like probe effects, nondeterminism, and complex communication patterns. In this paper, we describe the design and implementation of a real-time monitor for ARM processors which are frequently used for embedded applications. The focus of design is to help users debug real-time programs while minimizing the probe effect. Our monitor provides cross debugging features like down-loading from host, break-point based debugging features, and watch-point debugging features for real-time applications. We developed the debugger for ARM processor and debugger has been used for kernel program.

New methods for storing and retrieving semi-structured data such XML documents has been recently studied. In this paper, we concern querying of XML data stored in relational tables. We first describe a method of querying XML documents using Oracle XDK, especially XSQL Servlet. We then apply the method to XML documents describing book information.

SCM(Supply Chain Management) is concerned with the flow of product and information between the supply chain member organizations. Those organizations are suppliers, customers, producers, and service providers. They link together to acquire, purchase, manufacture, assemble, and distribute goods and services from suppliers to users. These flows are bidirectional. There(or is SCM specially important to integrate for efficient management. In this paper, is SCM based on XML(eXtensible Markup Language) studied.

Scheduling algorithm proposed in this paper is based on both DWRR(Dynamic Weighed Round Robin) method and QLT(Queue Length Threshold) method. The proposed scheduling algorithm guarantees delay property of realtime traffic, not considered in previous DWRR method, with serving realtime traffic preferentially by using RR(Round Robin) method which service each channel equally and QLT algorithm that is dynamic time priority method. Proposed cell scheduling algorithm may increase some complexity over conventional DWRR scheme because of delay priority based cell transmission method. However, due to reliable ABR service and congesition avoidance capacity, proposed scheduling algorithm has good performance over conventional DWRR scheme. Also, delay property based cell transmission method in proposed algorithm minimizes cell delay and requires less temporary buffer size

This paper describes the design of multiplier that receives two N-bit number and produces an N-bit product, with leading 0/l detector logic for an overflow prediction. A leading 0/l detector for two's input predict a scope of output. The part of partial products sum of N most-significant bits is exchanged for an overflow prediction. Therefore this multiplier requires less gates for the implementation about 45% than general multipliers.

최근 인터넷 이용이 증가하면서 원격 교육자료, 전자상거래용 자료 등 여러 이용자가 사용하는 자료를 하나의 서버가 아니라 분산된 형태로 나누어 관리하는 것이 통신량을 줄이기 위해 점차 중요시되고 있다. 특히 멀티미디어 자료는 데이터 량이 많으므로 이용자에게 가까운 서버에 분산 관리하는 것이 더욱 필요하다. 본 논문에서는 CORBA 표준을 따르는 미들웨어를 사용하여 인터넷으로 상호 연결된 임의의 시스템에 자료를 분산 관리할 수 있는 분산형 자료 관리 시스템 DDM(Distributed Data Manager)을 구축하였다. DDM은 크게 나누어 사용자 관리, 데이터 관리, 시스템 상태 관리 기능을 제공하며 단순히 분산 자료의 위치만을 관리하는 것이 아니라, 자료를 종류별 또는 용도별로 통합된 category를 관리한다. DDM은 분산 자료의 통합관리와 시스템 상태 관리를 위해서 주 에이전트(Primary Agent, 이하 PA)와, 이용자에게 직접 서비스를 제공하는 보조 에이전트(Local Agent, 이하 LA)를 중심으로 구성된다. DDM은 모든 자료에 대한 현재 상태를 관리하는데, 자료의 등록, 수정, 삭제 등의 요구가 보조 에이전트를 통하여 실시간으로 이루어지며 자료의 통합 관리와 백업 관리는 주 에이전트가 담당한다. 한편 DDM는 자바와 JDBC를 사용하여 구축함으로써 임의의 플랫폼과 데이터베이스를 이용할 수 있게 하였다.

Currently, the most important technology is the compression methods in the multimedia society. Audio files are rapidly propagated through internet. MP-3 is offered to CD tone quality in 128Kbps, but 64Kbps below tone quality is abruptly down and high bitrate. on the other hand, MPEG-II AAC (Advanced Audio Coding) is not compatible with MPEG-I, but AAC has a high compression ratio 1.4 better than MP-3. Especially, AAC has max. 7.1 channel and 96KHz sampling rate. In this paper, the perceptual model is dealt with 44.1KHz sampling rate for SMR(Signal to Masking Ratio)

As network resources have become faster and demands for multimedia service through network have increased, the demand for Media server system has increased. These kinds of media server solve their bottle neck problem of internal storage device by using parallel system which takes advantage of fast network resource. Many vendors have suggested each of their media server system to solve these problem radically, but most of them require major modification of infra component and additional drawback has added. For example, storage mechanism for specific media requires new file system which is totally different from traditional one, and algorithm for enhancing performance may not suit for traditional operating system environment. In this paper, we designed a parallel media server based on web interface of traditional system and implemented a program for media server. Implemented server system performs parallel processing through web interface without any modification of traditional system, and controls which is related to merging load by distributed data is charged only to client and control server and consequently load of storage server can be minimized. And also, data transfer protocol for streaming media includes Retransfer algorithm and client Admission control policy relevant to performance of whole system.

In this paper, we implement a video conference system, which connects remote PCs. Each system receives video and audio data from devices attached to a PC and compresses them by using CODECs. It sends the compressed data to the opponent. Compared to the existing ones, the proposed system can combine two video streams into a new one. Thus, additional video data can be transferred to the opponent during a call.

As Internet and WWW grow rapidly, the role of web servers is getting more important, and the number of users in a few popular site is also explosively increasing. Load balancing in clustered web server systems is important task to utilize whole system effectively, so dynamic load balancing is required to overcome the limit of static load balancing. In this paper, we propose two dynamic load balancing schemes, and analyzed load model and Performance improvement and also compare existing load balancing methods and IP filtering method. In case of load balancing with threshold, little extra traffic was required for better performance, but in case of load balancing with load weight, we found that the performance mainly depends on information exchange rate.

The EPRCA(Enhanced Proportional Rate Control Algorithm) is one of the explicit rate control algorithm recommended by the ATM forum. EPRCA algorithm is using only two threshold to buffer change is not considered for calculating ER value. In this paper to cover the shortage, two more variables are added for considering increase of buffer quantity. So calculating proper ER would avoid before getting into the congestion situation. To analysis the affect, EPRCA and proposed algorithm are compared through simulation by analysis of change of buffer length and ACR change

In this paper, we present new scheme for image coding which efficiently use the relationship between the properties of spatial image and its wavelet transform. Firstly an original image is decomposed into several layers by the wavelet transform, and simultaneously decomposed into 2$\^$n/ ${\times}$ 2$\^$n/ blocks. Each block is classified into 3 regions according to their property, i.e., low activity region(LAR), midrange activity region(MAR), high activity region(HAR). Secondly we are applied texture modeling technique to LAR, MAR and HAR are encoded by Stack-Run coding technique. Finally our scheme Is superior to the Zerotree method in both reconstructed image Quality and transmitted bit rates.

This study is to construct a general purpose library for a computer vision system development. The library includes many core algorithms required for computer vision systems such as image Processing algorithms, feature extracting methods, neural networks and etc. We have experimented the efficiency of the library by building a vehicle plate recognition system and the overall time and effort in development could be reduced to a certain extent.

Goal of this Paper is to design and develop core kernel components f3r single processor real-time system, which include real-time schedulers, synchronization mechanism, IPC, message passing, and clock & timer. The goal also contains the basic researches on dynamic load balancing and scheduling which provide mechanism for the distributed information processing and efficient resource sharing among various information appliances based on network.

In this paper, we proposed the high quality digital audio watermarking algorithm in the frequency domain. The spread spectrum technique is used to encrypted a stream of information by spreading the data as much of the frequency spectrum as possible. It's technique adapt well to data hiding in audio signal. We have used the perceptual model and MDCT/IMDCT for the high qualify digital audio watermarking. The proposed watermark algorithm preserved high quality audio data from watermark signal.

It is difficult to find web contents for visual disabilities in spite of rapid growing internet users, infrastructures and advance of computer technology. For the case of usenet newsgroup concerning disabilities, hit rate is even less than E-mail because of lack of accessibility, educational support and economical expense for internet. This paper is addressed to voice usenet newsgroup service by using telephone or web browser for visual disabilities without additional S/W support such as 775, usenet program and installation program, suggests a design method and an implementation example for it. Main features of it are easiness of man machine interface, popularity of access device such as telephone or web browser and independency of particular news server by using NNTP. This system supports general MIME format, is implemented for usenet server of Korean Social Worker's Community and will be implemented for Gomduri InfoNet BBS of Korean Society for Rehabilitation of Persons with Disabilities.

A RVLC(Reversible Variable Length Code) with an efficient table memory is proposed in this paper. In the conventional decoding methods, the weight of symbols and code values are used for the decoding table. These methods can be applied for Huffman decoding. In VLC decoding, many studies have been done for memory efficiency and decoding speed. We propose an improved table construction method for general VLC and RVLC decoding, which uses the transition number of bits within a symbol with an enhanced weight decomposition. In this method, tile table for RVLC decoding can be implemented with a smaller memory

Current Internet and Intranet developments are focusing on network multimedia services involving the transport of real-time multimedia streams. In this paper, we present a real-time MPEG2 Encoding/Decoding systems and a MPEG2 stream player running on client systems based on real-time transport protocol(RTP). The MPEG2 player has been implemented by using JMF and java applet. We have tested our prototype systems with one of possible multimedia sources, camcoder, to verify its functional correctness and measure its performance.

ATM network is considered as best suitable technology for multimedia service in various aspects such as bandwidth, capability of traffic expandability and so on. In this paper, we suggested a buffer management algorithm in ATM network to improve overall network performance with threshold and auxiliary buffer whose input consists of superpositiI on of voice and multimedia data traffic. To evaluate the proposed buffer management algorithm simulations are executed with four priorities, that is delay and loss priorities and the results are proved that network throughput is improved better than the existing partial buffer method.

In the Internet, multimedia data transfer is hard to guarantee the characteristics of the Real-Time because of the Best-Effort of the nature of IP, then additional mechanism is applied to multimedia application for real-time data. In this paper, we introduce the nature of multimedia and the necessary facility for real-time protocol. We propose protocol layer, which has necessary function above mentioned and offer the end-to-end transfer far real-time data. Also, the proposed protocol perform a next low operation: 1) a required information for QoS control by using Feedback mechanism is obtained from sender, 2) divided a transferred packet by delay and loss priority. 3) recognized the low service models, and 4) decided a bandwidth and QoS according to a network state

In this paper, we developed an outdoor augmented reality system which has remote real scenes acquisition ability. The real scenes acquisition system consists of Image acquisition system, tracking system, wireless data transceiver and power supply. The tracking system consists of Tans Vector and RT-20 which measures a position and attitude of the CCD camera that attached to the remote control helicopter. Wireless data transceiver system is utilized for data transmission of remote system that of attitude, position information, and real scenes data that acquired by the CCD camera. Maximum propelling power of remote control helicopter is 15Kg, so we used 7.2V li-ion cell as a power supply for system minimize. As the results of experiment, the developing system presented application possibility of remote information acquisition system such as construction simulation & estimation, broadcasting, tour guide.

This paper proposes a color-based image retrieval method using color adjacency and gradient. In proposed method, both the adjacency of different colors and gradient of a color in homogeneous region are considered as features of an image. The gradient, defined as the maximum distance along the direction with largest change of color, is computed for each pixel to determine whether the center color is similar or different to the neighboring colors. Therefore the problems caused by uniform quantization, which is popularly used at most existing retrieval, can be avoided. And furthermore, the storage of the feature is reduced by the proposed binary representation.

We propose the genetic algorithm to apply three kinds of FFT data flows to be considered the overhead for the data exchange between processors that have the multi-scheduling problem on parallel computer In the design of genetic algorithm, we propose the chromosome representation which can simply encode and decode a solution without any heuristic information, the evaluation function to be considered an efficiency of processor, and the genetic operator to inherit a superior gene from their parents. And we saw that the simulation result can verify better performance than the existing algorithm(BEA : binary exchange algorithm)in the face of execution time.

First, we give a linear-time algorithm to find the first overlap in an arbitrary binary word. Second, we implement the algorithm in the C language and show that the number of comparisons in this algorithm is less than 31n, where n$\geq$3 is the length of the input word.

IP Switching is a new routing technology proposed to improve the performance of IP routers. Flow classification is one of the key issues in IP Switching. To achieve better performance, flow classification should be matched to the varying IP traffic and an IP switch should make use of its hardware switching resources as fully as possible. This paper proposes an adaptive flow classification algorithm for IP Switching. By dynamically adjusting the values of its control parameters in response to the present usage of the hardware switching resources, this adaptive algorithm can efficiently match the varying IP traffic and thus improve the performance of an IP switch.

In this study, a new multimedia data converter is proposed. Also the PC interfacing met hod using the parallel port of is suggested. The image compression/decompression is based on the JPEG algorithm, which is widely used for an effective compression in the image processing industry. The suggested interfacing method is based on the IEEE1284 and IEEE1284.3 protocol, which is a standard in the PC's parallel port interface.

Recently, a feature extraction method based on decision boundary has been proposed for neural networks. The method is based on the fact that all the features necessary to achieve the same classification accuracy as in the original space can be obtained from the vectors normal to decision boundaries. However, the normal vector was estimated numerically. resulting in inaccurate estimation and a long computational time. In this paper. we propose a new method to calculate the normal vector analytically. Experiments show that the proposed method provides a better performance.

In this paper, we propose new features for describing texture images by using multi-channel Gabor filter and log-polar transform based on human visual system (HVS). Gabor features are extracted by the mean and standard deviation of energy in Gabor response, followed by Fourier series extension. Log-polar features are extracted by log-polar transform and projection. The proposed texture descriptor performs reasonably well with less number of features than other texture descriptors, which has been verified by experiments using some texture images of MPEG-7 data set.