The rotor flux level need be changed frequently for field weakening or power efficiency control. Motor inductances depend on rotor flux but not on machine temperature. On the other hand, rotor resistance varies greatly with the machine temperature. Motor parameters such a sinductances and rotor resistance should be known precisely in order to attain high dynamic performance of inductin motor. In this paper, efficient an dnovel identification algorithms for motor inductances and rotor resistance are presented. The rotor flux is changed. As the result, the slip frequency is varied. The identificatin algorithm for rotor resistance measures the varied slip frequency and alters the estimated rotor resistance. Then, the estimated value of rotor resistance will approach its real value. The proposed identification algorithms are computationally simple and have very small identification errors.

A numerous designs of PI controllers have been suggested to solve out trade-off between tracing and regulating problems. We constructed the PI controller system with two-degree-of-freedom that is more analytic and a better approach to a practical one. In the conventional H$_{2}$ design of optimal PI controllers, the cost function includes only the plant output terms due to the divergent problems. Since the platn input temr is not considered in PI controller design, occasionally, the plant input thends to be either very large or saturated. To solve the prior mentioned problems, we employed a mixed $H_2/H_{\infty}$ method that combines the H$_{2}$ design method to decide optimal parameters of PI controller and the $H_2/H_{\infty}$ design method to minimize the maximum amplitude of plant input. The calculation time of the H$_{infty}$ norm was considerably reduced by the simple scalar function obtained by the wiener-hopf factorization of non-scalar functions.

NUMA(Non-Uniform Memory Access)구조의 공유 메모리 다중처리기 시스템에서 참조 국지성의 활용은 병렬 처리의 성능에 큰 영항을 미친다. 본 논문에서는 운영체제가 참조 국지성을 관리하는데 도움을 주기위한 개선된 하드웨어 메모리 참조 카운터를 제시한다. 제신된 참조 카운터 방식에서는 기존의 참조 카운터들과는 달리 운영체제의 페이지 복사 정책을 다양한 메모리 참조 패턴에 적응시키기 위해 카운터의 값이 동적으로 그리고 주기적으로 조정된다. 우리는 실제 병렬 응용 프로그램들을 사용한 실행 구동형 시뮬레이션을 통해 제시된 "조정가능한 지연 카운터"가 이들의 성능에 미치는 영향을 평가하였다. 이 성능평가를 통해 '조정가능한 자연 카운터"를 이용한 메모리 복사 정책이 기존의 카운터를 이용한 정책보다 나은 성능을 보인다는 것과 시뮬레이션에 사용된 대부분의 병렬 응용 프로그램에 대해 고른 성능을 나타낸다는 것을 확인하였다.

A transputer is a powerful micro-processor, which can communicate with other processors with its own communication links and can be utilized into a powerful parallel computer system with easy way and low price. However, its usage has been limited to a few people because of its own programming language and complicated user interfaces. This paper presents programming tools to use a transputer system easily. It includes transputer integrated environment which has many developing tools, and transputer manager which can monitor and manage the system and developed programs on it. We also implemented a graphichs application program in order to test the feasibility and funcitonalities of programmers easy access to the transputer networks and visualize the performance status of each processor so that they can write efficient parallel programming codes.

In this paper, the convergence behavior of timing phase when an adaptive partial response equalizer and decision-directed type of a PLL work together in a digital recording channel is described. The phenomena of getting biased in timing phase when the convergence parameter of an adaptive partial response equalizer and timing recovery constant of a PLL are not selected properly is introduced. The phenomena, occurring due to perturbation of timing phase, are analyzed, by computer simulation and the region of ocnvergence for timing phase is discussed. Also, a method to overcome the phenomena using a variable step-size parameter is described.

We propose a 3rd order blind equalizer that incorporates a new transform method using either square root operation ($\sqrt{x]$) or reciprocal operation (1/x) in order to transform symmetric distribution of PAM signals at the transmitter, to asymmetric one. At the receiver, either the square operation or the reciprocal operation is needed to recover the asymmetrically transformed signals to the original ones after eualization. The reslts of the computer simulation, using the new method are better than the existing transform method using natural logarithm operation by the maximum of 8 dB in MSE. In addition, as the skewness of the asymmetrically transformed distribution has small values, the performances are improved.

In this paper, we propose predictive motion estimatin algorithm which can predict motion without additional side information considering spatio-tempral correlatio of motion vector. This method performs motion prediction of current block using correlation of the motion vector for two spatially adjacent blocks and a temporally adjacent block. Form predicted motion, the position of searhc area is determined. Then in this searhc area, we estimate motion vector of current block using block matching algoirthm. Considering spatial an temporal correlation of motion vector, the proposed method can predict motion precisely much more. Especially when the motion of objects is rapid, this method can estimate motion more precisely without reducing block size or increasing search area. Futhrmore, the proposed method has computation time the same as conventional block matching algorithm. And as it predicts motion from adjacent blocks, it does not require additional side information for adjacent block. Computer simulation results show that motion estimation of proposed method is more precise than that of conventioanl method.

In this paper, based on the constrained optimizatin technique and direct recovery method, we proesent a shape form shading (SFS) algorithm to recover a 3-D shape form an image. More specifically, we first employ the membrane model for a smoothness constraint to revoer a 3-D shape coarsely. We then compute the surface height directly to reduce the shape distortion due to a regularization term. In our approach, we can obtain a stable and accurate solution by the application of these two steps. Several simulation results on various images are provided and discussed in this paper and they show that the proposed algorithm extracts the 3-D information accurately and efficiently.

In this paper, an image coding technique using fractal interpolation is proposed. Similar to the conventional methods, an image is partitioned into blocks and each block is coded independently. However, an interpolation point is ahsared by its neighboring blocks. This means that each block can use all its interpolation points with minimal increase of new data. For a simple implementation, triangular blocks are used instead of square blocks and new coefficients are difined. Data obtained in the encoding process hav estatistical characteristics suitable sfor entropy coding, an dthus arithmetic coding is perfomred for improving the compression efficiency. The results of the proposed coder in comparison with those of a conventional coder show that the interpolation method reduces block effect caused by a memoryless block coder, especially at low bit rates. This improvement is due to sharing of information between adjacent blocks. Moreover, th enumber of iteration required in ecoding process is reduced since more information is used to decode each block.

This paper presents a new mehtod with which we can improve the efficiency (both of speed and precision) of FPN$^{[1]}$. Our method makes use of the basic propositions smong the conditional propositions of a goal proposition. The basic propositions are represented as ource vertices on an FPN. Therefore, we introduce an efficient algebraic algorithm which finds source vertices of FPN.

A new design scheme of a binary decision tree is proposed. In this scheme a binary decision tree is constructed by using genetic algorithm and FCM algorithm. At each node optimal or near-optimal feature subset is selected which optimizes fitness function in genetic algorithm. The fitness function is inversely proportional to classification error, balance between cluster, number of feature used. The binary strings in genetic algorithm determine the feature subset and classification results - error, balance - form fuzzy partition matrix affect reproduction of next genratin. The proposed design scheme is applied to the tire tread patterns and handwriteen alphabetic characters. Experimental results show the usefulness of the proposed scheme.

A content-based image retrieval system is designed and implemetned using the color featurees which are histogram intersection and color pairs. The preprocessor for the image retrieval manage linearly the existing HSI(hue, saturation, saturation, intensity). Hue and intensity histogram thresholding for each color attribute is performed to split the chromatic and achromatic regions respectively. Grouping te indexes produced by the histogram intersection is used to save the retrieval times. Each image is divided into the cells of 32$\times$32 pixels, and color pairs are used to represent the query during retrievals. The recall/precision of histogram intersection is 0.621/0.663 and recall/precision of color pairs is 0.438/0.536. And recall/precision of proposed method is 0.765/0.775/. It is shown that the proposed method using histogram intersection and color pairs improves the retrieval rates.

The proposed method uses a signal of the smae frequency as the input modulating carrier frequency and of a different phase. This signal is generated in the digital automatic frequency control part to decide the input color demodulated signal. And the phase error from the burst signal is calculated. The calculated phase error is utilized to rmove the phase error contained inthe demodulated color signal. In this paper, digital signal processing of automatic color control is proposed for VCR system campatible with both NTSC and PAL TV systems.

In this paper, a digital signal processing method of AFC (automatic frequency control) is proposed for a home use VCR system. The proposed method has the ability of frequency tracking of a wide range. Implemented with digital circuits, the system is to be used in a digital video system and saves the cost of a hardware compared with a conventional analog automatic frequency control method using several PLL's in case of making home use VCR systems compatible with several TV systems.

In this paper, we analyze the traffic characteristics in microcell and macrocell overlaid PCS system which process voice and dta calls separately each others. in this system, data calls are delayed in queue when all of channels are occupied, while voice calls are bolcked in that case. For this, we calculated inter-microcell handoff area dwelling time distribution and handoff area dwelling time distribution between microcell and macrocell. We analyze traffic performance using this results. We used M/M/C/K model, and analyzed traffic performance of macrocell with handoff area variation of microcell.

In this paper, we designed binary phase grating with SA (simulated annealing) algorithm for image generation, and used LCTV(liquid crystal television) as phase modulator. We generated optical image using LCTV which controlled by personal computer in real time. Many parameters in SA algorithm are determined by average deviation of the cost function, not experimental method. It was confirmed that the LCTV could be used as a phase modulator through mach-zehnder interferometering experiment. The various optical images are generated in real itme by fourier transforming of the phase gratings on LCTV which is controlled by personal computer and they can be used for optical interconnection in a neural network and addressing information in a volume hologram etc.

In this paper, we propose VLSI architectures for the full search block matching algorithm (FS BMA) and two-stage BMA using integral projections that reduce greatly computational complexity with its performance comparable to that of the FS BMA. The proposed VLSI architectures are faster than the conventional ones with lower hardware complexity. Also the proposed architectures of the FS BMA and two-stage BMA are modeled in VHDL and simulated to show their functional validity.

In this thesis, to cope with severe intersymbol interference and nonlinear distotions of the digital magnetic recording channel a neural decision feedback equalizer (NDFE) and an adaptive neural equalizer (NE) are applied. The digital magnetic recording channels with various recording densities and different types of the nonlinear distortions are considered. The computer simulation shows that as the nonlinear distortion is increased, the neural equalizers (NdFE, NE) have advantages of approximately 2-4 dB in signal to noise ratio (SNR) over the onventional eualizers to reach sme bit error rate and, a sthe recording density is increased, 1~5 dB of SNR improvement are also gained. Especially the NdFE gives a superior performance over the other equalizers when there is a severe nonlinear distortion in the digital magnetic channel.

In this paper, we present the modified CMA (constant modulus algorithm) and LMS (least mean square) algorithms for digiCipher system with reduced hardware cost, in which the pipelined architecture is employed. They yield the performance comparable to that using floating-point operations. We show the effecstiveness of the proposed architecture through the implementation results using VHDL.

A multiple target tracking (MTT) problem is to track a number of targets in clusttered environment, where measurements may contain uncertainties of measurement origin due to clutter, missed detection, or other targets, as well as measurement noise errors. Hence, an MTT filter should be introduced to resolve this problem. In this paper, a neural network is rpoposed as an MTT filter.

A PSRI (position, scale, and rotation invariant ) feature extraction and automatic target recognition system using a cooperative network and an MLP is proposed. We can extract position invarient features by obtaining the target center using the projection and the moment in preprocessing stage. The scale and rotation invariant features are extracted from the contour projection of the number of edge pixels on each of the concentric circles, which is input to the cooperative network. By extracting the representative PSRI features form the features and their differentiations using max-net and min-net, we can rdduce the number of input neurons of the MLP, and make the resulted automatic target recognition system less sensitive to input variances. Experiments are conduted on various complex images which are shifted, rotated, or scaled, and the results show that the proposed system is very efficient for PSRI feature extractions and automatic target recognitions.

This paper presents a decentralized control scheme for interconnected systems with unmodeled nonlinearities and interactions using a neural coordinator. The interactions due to the interconnection and the unmodeled nonlinearity associated with each subsystem are represented by the deviations from linearized states of decomposed subsystems. the decentralized controller is composed of local controllers and a neural coordinator. The local controller for each subsystem is derived from linearized local system parameters y linear optimal control theory. the neural cooridnator generates a corrective control signal to cancel the effect of deviation sthrough the backpropagation learning with the rrors obtained form the difference of the local system outputs and reference model outputs. the reference model consists of the part of local system without deviations. The effectiveness of the proposed control scheme is demonstrated by simulation studies.