This paper discusses the novel approach to solve the VLSI channel problem. Routing is one of the major tasks in the physical design of VLSI. Terminals that belong to the same net are connected together subject to a set of routing constraints. With new performance requirements for the design. routing constraints such as crosstalk and RC delay between interconnections are becoming increasingly dominant in submicron regimes. Crosstalk is delay due to coupled capacitance and RC delay is caused resistance and capacitance. Hence, the proposed algorithm based on Genetic Algorithms optimizes both physical constraints (number of tracks) and electrical constrains (crosstalk and RC delay). For selection control, which is one of the genetic operations, new objective functions for minimum crosstalk and RC delay are introduced in addition to the usual objective function based on number of tracks. The experimental results prove that the proposed algorithms is effective for reducing crosstalk and RC delay, as well as number of tracks.

抄録全体を表示

It is useful, for objectivity and effectiveness of the power transformer impulse test, to computerize its diagnosis procedure and fault location identification process. It is possible to recognize measured current patterns using Kohonen's self-organizing map which only requires a learning process of the relationship between fault locations and their transfer characteristics. The transfer function method which computes transfer characteristics via FFT (Fast Fourier Transform) is used for preprocessing the fault characteristics. An example of applying the simple mathematical model of a transformer for results simulated using the developed Matlab program is reported.

抄録全体を表示

A linear discrete-time system represented by delta-operator is known to have an advantage of accuracy in numerical calculations over usual shift-operator systems. Analysis of such a system is a topic of recent times. In this paper, we are interested in stability analysis for delta-operator systems with parametric uncertainties represented by interval polynomials. Though the extreme point results hold for the stability of such polynomials, computational cost becomes markedly enormous when the degree of the polynomials increases. We propose a new stability analysis method for the systems using stability margin in order to reduce the amount of work for stability analysis. We check if a hyperbox of an interval polynomial is contained in the stability region in the coefficient space. To do this, we propose ‘directional stability radius’ as a stability margin estimater. It is a stability radius where coeffient perturbations are supposed to be in certain constrained directions. The devised stability analysis method is to test which vertexes of the hyperbox are covered by certain stability hyperballs with the directional stability radius. By numerical examples, we show that the proposed method is more efficient than brute-force check of every vertex polynomial.

抄録全体を表示

The effectiveness of neural networks is discussed for nonlinear systems. Especially, some control schemes by using the neural network have been proposed. The conventional neural-net based control schemes can be classified into the two groups. The one is that control input is directly calculated by the neural network. As the control input is given by the output of the neural network, it is easy to implement for the controlled object. However, it is difficult to describe the structure of the controller as the transfer function. The other is that control parameters are calculated by the neural network, and this control input can be calculated by these control parameters. Although this scheme makes easy to grasp the physical meanings of control parameters, the properties of controlled object can not be directly understood. In this paper, a parameter estimation scheme for nonlinear systems is proposed by using a neural network. Furthermore, a design method of the control systein is derived by minimizing a cost function. This control input is calculated by using the estimated parameters. Finally, the effectiveness of the proposed scheme is numerically evaluated.

抄録全体を表示

The chaotic neural network is one of the most effective neural models as a solver of combinatorial optimization problems. Hardware implemention of the model contributes to a fast solver of the problems. Although FPGA (Field Programmable Gate Array) hardware imprementations of the model are really attractive because they can be easily designed and modified by using a hardware description language, it is difficult to implement the chaotic neural network because output of the neuron is real value. Implementation of real value on digital circuits needs huge area on the chip and few neuron can be contained in a chip. In this paper, output of the neuron in the chaotic neural network is binarized for FPGA hardware implementation. It is applied to N-Queen problem, and from numerical experiments it is confirmed that the neural network with binary output is more effective than the original chaotic neural network. A prototype of the proposed model is designed by VHDL and implemented on the FPGA Xilinx XC4028XL.

抄録全体を表示

In this letter, a cladding-layer-loaded LiNbO₃-waveguide optical amplifier is investigated from the point of the view of improving the gain in the light-amplification. Loading a cladding-layer, the modal field in the waveguide is moved toward the guide-surface, and the mismatch between the distribution of the doped rare-earth atom and the modal field is compensated. As a result, the gain mainly dependent on the interaction integral between the dopant distribution and light field can be expected to improve. Simulation results on the dependence of the gain on the cladding-layer thickness are demonstrated.

抄録全体を表示

Spontaneous neuronal burst activities were optically recorded in the neonatal rat brainstem slice preparation (1-3days after birth) stained with a voltage sensitive dye (RH-795). Dledullarv spontaneous bursts were continuously monitored from the central cut-end of the XIIth cranial nerve in the standard artificial cerebro-spinal fluid. During a respiratory cycle including one bursting phase and one silent phase (approximately 10s), a, fluorescence change corresponding to spontaneous XIIth cranial nerve activity was microscopically observed both in the hypoglossal nucleus and the ventrolateral medulla (VLM) in which the localization of many respiratory neurons had been demonstrated in physiological studies. The present results show the feasibility of using optical recording for the analysis of spontaneous neuron activity.

抄録全体を表示

In order to measure biomagnetic fields such as magnetoencephalogram (MEG), magnetically shielded room (MSR) with multi layers of μ-metal is usually used to reduce environmental magnetic field noise. However, such heavy-shielded room is not suitable for magnetocardiogram (MCG) measurement system from the requirements of low cost, lightweight, and small space.
We have studied an environmental noise reduction scheme for MCG application. Our noise reduction system consists of a prefabricated MSR of single μ-metal layer, a noise detection coil, and compensation coils. The following results were obtained for the suppression of environmental noise: 30 dB at 0.5-10 Hz and 30-50 dB at 10-100 Hz.
In this paper, the compensation set-ups are described and experimental results are presented and discussed.

抄録全体を表示

This paper describes the basic characteristics of a new method for recording tone images by applying zigzag scanning to ink flight in the electrostatically extracted ink jet method using dielectric ink.
Zigzag scanning is a method for recording tone images by deflecting flying ink jet right and left with the nozzle tip as the starting point. Behavior of zigzag scanning was examined by varying applied voltage, ink pressure, frequency, and positional relationship between the opposing electrodes. The results obtained revealed the following basic characteristics. The flying ink in the zigzag scanning showed dependency on the applied voltage and ink pressure, and the higher the voltage and the lower the ink pressure within a given range, stability was obtained. Also the response to the frequency was about 3 kHz. An examination was further made on application of zigzag scanning to recording of tone images on the basis of these characteristics. As a result, the zigzag scanning, in addition to the liquid thread in the ink flight and minute corpuscle diameters which are the features of this ink jet method, was found to be an effective method for achieving high-quality in tone image recording.

抄録全体を表示

Formatting process of earth observation satellite downlink data is often used by custom made hardware which designed for one satellite or a sensor. However, in the near future lunch of various kinds of multi sensor satellite such as ADEOS-2 are planned. If we make hardware for each sensor for format processing, it the system, and operation becomes very complicated. It also causes degradation of reliability of the format processing system. This is a very big problem. We have developed a software based downlink data formatter which solved these problems. The hardware dependency was reduced and made multiusable high speed processing was realized.

抄録全体を表示

A dedicated wireless data collection system to monitor city lifelines, i.e., water, electricity and gas supply, during- and post-eismic hazard has been investigated using an experimental system. The system employed a 2.4GHz TD-CDMA technique as a new application of CDMA. The TD-CDMA system collected data from 256 household terminals in a cell in 10.24sec with a frame error rate (FER) at 2×10^-3 When a wireless LAN (W-LAN) was brought into operation in the same frequency band as the experimental system, interference occurred. A method using an attenuator placed in the input port of the base station receiver in conjunction with CDMA power control has been developed to mitigate the interference in a CDMA fixed station. While W-LAN operating, the interference power increased from -106dBm to -87dBm, and the FER deteriorated to 3×10^-2 or more. When attenuation in the range of 3dB - 11dB was placed in the input port of the receiver, CINR (Carrier-to-Interference-Noise Ratio) improved from -13dB to -10dB or more, provided that the power control function remained effective. As a result, the FER recovered to a level at 1×10^-3. The experiment has demonstrated the effectiveness of the interference mitigation technique.

抄録全体を表示

In the echo cancellation, the double-talk situation when both the near-end signal and far-end signal are presented degrades the performance of echo cancellation with the LMS or the NLMS algorithms. In order to obtain an acceptable performance even in the presence of the double-talk, the authors proposed the expanded correlation least mean squares (ECLMS) algorithm but this algorithm does not give a fast conver-gence characteristics still. The purpose of this paper is to derive an orthogonal ECLMS algorithm with the lattice strucure in order to speed up the convergence characteristics and to reduce the computational load for the adaptation. The validity of this algorithm is evaluated by computer simulations in the echo cancelling problem.

抄録全体を表示

In this paper, a robust personal identification system which is adaptable for either rotated or tilted faces is presented. In the proposed method, the 25 facial images which are taken with different angles are registered in the database as reference images. The similarity of an input image is first examined between all registered images and the three facial images, which are similar to the input image, are then extracted as the candidate facial images. The isodensity maps, which is expected to have the maximum similarity to the input image, is created by the interpolation between two candidate images and the input facial image is judged identical or not with the interpolated images. Computer simulation using 40 people shows 94.6% correct discrimination accuracy rate. From the experimental results, the prospects of using this method are very encouraging.

抄録全体を表示

This paper describes a new model of multi-valued immune network based on biological immune response network. The model of multi-valued immune network is formulated based on the analogy with the interaction between B cells and T cells in immune system. The model has a property that resembles immune response quite well. The immunity of the network is simulated and makes several experimentally testable predictions. Simulation results are given to a letter recognition application of the network and compared with binary ones. The simulations show that. beside the advantages of less categories. improved memory pattern. good recognitions rate and good memory capacity. the multi-valued immune network produces a stronger noise imununity than binary one.

抄録全体を表示

The purpose of this study is to detect accurately the target in the image which is measured by the Imaging Laser Radar System using the high power LD. The Imaging Laser Radar System is the active imaging system, which transmits the laser beams and detects the reflected light. And, the Imaging Laser Radar System can obtain the range image in addition to the intensity image, simultaneously. In this paper, we research and develop the Imaging Laser Radar System. Firstly, we construct the Imaging Laser Radar System to measure the target at a distance of 50m from system. And, we obtain the intensity image and the range image, simultaneously. Secondly, we discuss the relationship between the target detection probability and the number of false alarms of the binary coded image. Then, we show that the range image has the higher detection probability and lower number of false alarms than the intensity image. According to the optical properties, the intensity and range image can simply be merged to obtain 3-D image using the data fusion algorithm. Thirdly, we discuss the relationship between the target detection probability and the number of false alarms of the binary coded 3-D image. Then, we show that the 3-D image has lower number of false alarms and higher detection probability than the range image.

抄録全体を表示

In this paper we consider the function approximation when synthesizing an automatic choosing control of structure-specified type for nonlinear systems. Nonlinear terms of the system are piecewisely linearized by Taylor expansion so as to design the linear optimal controls. These controls are smoothly united into a single nonlinear feedback control by the sigmoid type automatic choosing functions (STACF). Performance of this control greatly depends on the approximation of the given system, so its error bounds are developed and investigated under various conditions. Using one of the error bounds. the parameters such as the optimal number of the STACF are evaluated from suboptimal point of view. Examples are provided to verify the tightness of these bounds.

抄録全体を表示

A method for robust model following control of nonminimum phase discrete-time systems using approximate inverse systems is presented. This paper uses an approximate inverse system applicable to the nonminimum phase discrete-time system which assures that the system input and output remain bounded at any time. Furthermore, this scheme can robustly control the nominal model in the presence of unmodeled dynamics and achieve the desired model following simultaneously. Furthermore, the sufficient condition for stabilizing the nominal model in the presence of unmodeled dynamics is derived and the existence of bounds for all signals is proved. Finally, computer simulation results are presented to show the effectiveness of the proposed method.

抄録全体を表示

A new method has been developed for the optimal radial network configuration using Genetic Algorithm. Network reconfiguration problem is one of the combinatorial problems. Therefore combination of search space becomes increases exponentially according to the number of switches. The searching space of the developed GA is reduced exponentially compared to that of conventional GA methods. Firstly (k-n+1) loops are decided in a network which contains k branches and n nodes when all section switches are closed. Every branch must be a member of loops by one or two times to make all radial network expressible and minimize searching space. Each loop is assigned serial number. Then in chromosome each number of open switch is lined up with the order of loop number. This expression reduces infeasible lethal genes after crossover or mutation operation.
The developed method has been applied to both reconfiguration problem with current constraint and realistic loss minimization problem. For the former problem, 4×5 mesh network, the optimal network is generated for all 20 cases. And for the latter problem, 112 branches and 102 nodes network, the transmission loss of the optimal generated network is within 8.6% compared to the lower bound (loop network). These results show effectiveness of the developed method.

抄録全体を表示

An extremely low phase noise oscillators realizes by oscillator with highly output level. If we want to measure a phase noise in such these oscillator, it is a necessary to modify output level as correspondent to dynamic range of measurement system. However, we have known nothing how many the effect exists.
This paper discusses on the effect of level adjustment, and examine on an evaluation of phase noise in extremely low phase noise oscillators.

抄録全体を表示

The excitation of solid state and dye lasers requires a number of flashlamps. We developed driving circuits with multiple flashlamps connected both in series and in parallel with a single spark gap switch. The design and performance characteristics of the proposed circuits are discussed.

抄録全体を表示

Boundary extraction is the partition of an image into a set of nonoverlapping regions whose union is the entire image. The image is decomposed into meaningful parts which are uniform with respect to certain characteristics. We propose a methodology for evaluating medical boundary extraction algorithm. In this case, the boundary extraction algorithm can be evaluated by integration of separability and gradient, which can be calculated by linear discriuninant analysis. We illustrate the use of this methodology on different applications in ultrasound imaging.

抄録全体を表示