First, a number of issues pertaining to analog VLSI implementation of Backpropagation (BP) and Deterministic Boltzmann Machine (DBM) learning algorithms are clarified. According to the results from software simulation, a mismatch between the activation function and derivative generated by independent circuits degrades the BP learning performance. The perfomance can be improved, however, by adjusting the gain of the activation function used to obtain the derivative, irrespective of the original activation function. Calculation errors embedded in the circuits also degrade the learning preformance. BP learning is sensitive to offset errors in multiplication in the learning process, and DBM learning is sensitive to asymmetry between the weight increment and decrement processes. Next, an analog VLSI architecture for implementing the algorithms using common building block circuits is proposed. The evaluation results of test chips confirm that synaptic weights can be updated up to 1 MHz and that a resolution exceeding 14 bits can be attained. The test chips successfully perform XOR learning using each algorithm.

A new model for a computer simulation of RF capacitive type hyperthermia has been developed by taking account of the following points. Blood flow is usually determined by many physiological parameters, but is regarded as a function of only blood temperature under some conditions. The temperature dependence of blood flow of tumors and normal tissues is assumed by referring the data obtained by Song et al. and Tanaka. The blood temperature which is elevated by externally applied power significantly affects temperatures of the body and the tumors. The transport of heat from the body surface is studied by considering air convection. These points are examined by experiments on a computer with simple phantom models and real patients. The results of simulation on the patient have shown a good agreement with clinical inspection based on CT images and a temperature of the stomach.

This paper presents an automatic creation method of 3D facial models which are needed for facial image generation by 3D computer graphics. A 3D facial model of a specific person is obtained from just the front and side view images without any human operation. The method has two parts; feature extraction and generic model modification. In the feature extraction part, the regions or edges which express the facial features such as eyes, nose, mouth or chin outline are extracted from the front and side view images. A generic head model is then modified based on the position and shape of the extracted facial features in the generic model modification part. As a result, a 3D model for persons is obtained. By using the specific model and the front and side view images, texture-mapped facial images can be generated easily.

New methods are needed for accessing very large information services. This paper proposes the use of a user model neural network to allow better access to a news service. The network is constructed on the basis of articles read, and articles marked as rejected. It adapts over time to better represent the user's interests and rank the articles supplied by the news service. Using an augmented keyword search we can also search for articles using keywords in conjunction with the user model neural network. Trials of the system in a USENET news environment show promising results for the use of this approach in information retrieval.

This paper presents a new fast fault simulation algorithm using a content addressable memory, which deals with zero-delay fault simulation of gate-level synchronous sequential circuits. The computation time of fault simulation for a single vector under the single stuck-at fault model is O(n2) for all the existing fault simulation algorithms on a sequential computers. The new algorithm attempts to reduce the computation time by processing many faults at a time by utilizing a property that a content addressable memory can be regarded as an SIMD type parallel computation machine. According to theoretical estimation, the speed performance of a simulator based on the proposed algorithm is equivalent to a fast fault simulator implemented on a vector supercomputer for a circuit of about 2400 gates.

By close analogy of annealing for solids, we devise a new algorithm, called APS, for the time evolution of both the state and the synapses of the Hopfield's neural network. Through constrainedly random perturbation of the synapses of the network, the evolution of the state will ignore the tremendous number of small minima and reach a good minimum. The synapses resemble the microstructure of a network. This new algorithm anneals the microstructure of the network through a thermal controlled process. And the algorithm allows us to obtain a good minimum of the Hopfield's model efficiently. We show the potential of this approach for optimization problems by applying it to the will-known traveling salesman problem. The performance of this new algorithm has been supported by many computer simulations.

The effect of the externally reflected light on the mode partition characteristics of 1.3 µm Fabry-Perot laser diodes is studied experimentally and numerically. It is observed that the k-value increases monotonically with the DC bias current and the external reflection coefficient. Based on these experimental results, a numerical model to study the mode partition characteristics of laser diodes in the presence of external reflections is developed. The results calculated using this model agree well with the experimental ones. It is found that the mode partition noise is mainly caused by the interference between the light in the laser diode and the reflected light, and also by the fluctuations of the induced emission and absorption. In the time domain, their contribution to the mode partition noise is almost localized in the time region within 0.1nsec at the time when the optical pulse turns on.

Information technology equipment connected to telecommunications line can be a source of electromagnetic interference. Two sources of interference have been under evaluation. One is the digital pulses in the switching regulator and the clock oscillator, and the other is the signal's common mode voltage. In this paper, the interference-inducing mechanism for the signal's common mode voltage and a method for measuring the interference are described. An equivalent circuit representing both the equipment and the line is derived on the basis of the interference-inducing model. A method for estimating the signal's common mode voltage from the differential mode voltage and the line unbalance is obtained using the equivalent circuit. It is confirmed that the level difference between the estimated and the measured common mode level is less than 3dB.

Serious failures of the latest electronic equipments occur easily due to electrostatic discharge (ESD) , which can be caused frequently by the electrification phenomena of human-body walking on the floor. The number of the above damaging incidents has significantly been increasing with an increased use of integrated semiconductor elements with lower operation power. The most effective measures against the ESD consist in preventive ones, which are to obtain dynamic behaviors of the electric charge before the ESD happens, thereby preventing the charge accumulation. From this point of view, this paper describes new approaches for measurement of the static electricity directed toward preventing the ESD. First, a two-dimensional measurement method for visualizing charge distributions is described. This principle is based on visualizing the potential distribution induced in the array electrodes from the electrostatic fields. For showing usefulness of the visualization measurement, a prototype was built and attempts were made on the visualizations for the static electricity distributions of charged bodies. Second, a potential calculation of the human body charged by walking on the floor is described. A model was shown for analyzing the human-body potential on the floor, and the theoretical equation for describing the potential attenuation process was derived in the closed form in the Laplacian transformation domain. In order to obtain the typical half-life of the human-body potential, numerical computations were performed using a reverse Laplacian transformation. The experiments were also conducted for confirming the validity of the computed results. Finally, a new method is described for estimating dynamic behaviors of the occurrence charges of the human body electrified by walking-motions. Statistical measurements of the charges and potentials were made for the fundamental walking-motions specified here. The pace transitions of the potentials due to continuous walking and stepping were also measured and their results were explained from the electrification properties for the fundamental walking-motions.

A Discrete Event System (DES) is a system that is modeled by a finite automaton. A Cooperating Discrete Event System (CDES) is a distributed system which consists of several local DESs which are synchronized with each other to accomplish its own goal. This paper describes the automatic synthesis of a CDES from a modular temporal logic specification. First, MPTS (Modular Practical Temporal Specification language) is proposed in which the new features (modular structure and domain specification) are appended to temporal logic. To overcome the "state explosion problem", which occurs in generating a global automaton in former synthesis methods using temporal logic, a compositional synthesis is proposed where automata are reduced at every composition step.

LOTOS is a language developed within ISO for the formal description of communication protocols and distributed systems. In LOTOS, requirements for a distributed system are called a "service specification". Each node exchanges synchronization messages to ensure the temporal ordering for the execution of events in a service specification. The actions of each node are described as a "protocol specification". This paper gives a survey for a method to derive protocol specifications from a service specification written in a LOTOS based language. In order to derive the protocol specifications, we make the syntax tree of a given service specification and give some attributes for each node in the tree. The protocol specifications are derived automatically by evaluating these attributes. The derived protocol specifications satisfy the given service specification. We also explain a LOTOS simulator for the execution of derived protocol specifications. The related works are also summarized.

It is sometimes required to change the frequency characteristics of a digital filter during its operation. In this paper a new synthesis of variable even-order IIR digital filters is proposed. The cut-off frequency of the filter can be changed by a single parameter. The fundamental filter structure is a cascade of second-order sections. The multiplier coefficients of each section are determined by using the Taylor series expansion of the lowpass to lowpass frequency transformation. For this method any second-order section can be used as a prototype, but here in this paper only the direct form and the lattice form are described. Unlike the conventional method, any transfer functions can be used for the proposed method. Finally a designed example shows that the proposed filter has wider tuning range than the conventional filter, and the advantage of the proposed filters is confirmed.

It has been recognized that there exist some disparities between properties of continuous control systems and those of discrete ones which are obtained from their continuous counterparts by use of a sampler and zero order hold. This still remains true even if the sampling rate becomes fast enough and sometimes causes unfavorable effects in control systems design. To reconcile with this conflict, use of delta operator has been proposed in place of z-operator recently. This note formulates a delta domain Lyapunov matrix equation and shows that the equation actually mediates the discrete Lyapunov equation and its continuous counterpart.

This paper investigates the compositionality of operational models for concurrency induced by labeled transition systems (LTS's). These models are defined on the basis of a metric domain first introduced by de Bakker and Zucker; the domain is a complete metric space consisting of tree-like structures called processes. Transition system specifications (TSS's) define LTS's; the set of states of such a LTS A is the set of terms generated by a signature Σ. For the syntactical operators F contained in Σ, semantic operations (on processes) associated with F are derived from the TSS S by which A is defined, provided that S satisfies certain syntactical restrictions. By means of these operations, the compositionality of the operational model induced by A is established. A similar result was obtained by Rutten from TTS's which define finitely branching LTS's. The main contribution of this paper is generalization of Rutten's result to be applicable to TSS's which are based on applicative languages including recursion, parameterized statements, and value passing, and which define infinitely branching LTS's. A version of typed λ-calculus incorporating µ-notation is employed as a formalism for treating recursion, parameterized statements, and value-passing. Infinitely branching LTS's are needed to treat programming languages including value passing such as CCS.

For the efficient circuit simulation by the direct method, network tearing and latency techniques have been studied. This letter describes a circuit simulator SPLIT with hierarchical decomposition and latency. The block size of the latent subcircuit can be determined dynamically in SPLIT. We apply SPLIT to the MOS circuit simulation and verify its availability.

An automatic optimization system for semiconductor devices has been built-up by fully interfacing an optimizer and a device-analysis code supplemented with sensitivity analysis. The device-analysis code is thought of as a part of a pipeline of simulators. The latters are regarded as subprocesses by the optimizer, which controls their I/O stream. The action of the pipeline is iterated until the optimum set of design parameters is determined. An important feature of the system is that all the derivatives required in the sensitivity analysis are calculated analytically, this providing a substantial improvement in both the numerical accuracy and computational efficiency, and making the scheme attractive from the application standpoint. A few examples of optimization of MOS devices are shown and the performance is reported, indicating that a system of this kind can usefully be exploited in a design environment.

We have developed an efficient general-purpose two-dimensional device simulation system which consists of a solver, and pre- and post-processors. This system can easily handle any complicated device having a non-rectangular shape. It can also be applied to compound semiconductor devices with heterojunctions, including optical devices such as laser diodes. In order to handle any device, a new program for construction of device geometry is developed as a preprocessor. It has an efficient graphic interface to reduce the time required to input data for simulations, which is a very time consuming task for complicated devices. A new efficient data structure representing device geometry is introduced in the program. During postprocessing, any physical quantity can be displayed on the multi-window screen. In addition, a general-purpose solver for basic semiconductor equations is implemented in the system. Using this system, any device can be successfully analyzed in a unified manner and the turn-around time for the simulation is significantly reduced.

A Monte Carlo calculation is performed to examine the transport coefficients of the electron gas under an inhomogeneous electric field. The expressions constructed from the M. C. results are then incorporated into the hydrodynamic formulation to calculate the internal characteristics of a silicon BJT device. The calculated results agree well with the Monte Carlo prediction.

In this paper, a "hydrodynamic" version of the three-dimensional code HFIELDS-3D is used to achieve a detailed knowledge on the distribution of the substrate current inside a recessed-oxide MOSFET. The physical model features a temperature-dependent formulation of the impact-ionization rate, allowing non-local effects to be accounted for. The discretization strategy relies on the Box Integration scheme and uses suitable generalizations of the Scharfetter-Gummel technique for the energy-balance equation. The simulation results show that the narrow-channel effect has a different impact on drain and substrate currents. Further three-dimensional effects, such as the extra heating of the carriers at the channel edge, are demonstrated.

Theoretical network analysis of a network constructed of "Interconnectable Star Couplers" whose all diagonal elements of transmission matrix are zero is investigated. Under certain connection rules, Interconnectable Star Coupler can be connected each other without oscillation and ghost formation. The rules are: (1) Network should not contain any loop. (2)Only single port pair should be connected between neighbor star couplers. (3)Network shold not contain any usual star coulpler. "Coupled Star Network", which is constructed under the connection rules, is able to form Cascade Star topology, Stratified Star topology and their mixed topology. It is shown that the Coupled Star Network is equivalent to a large Interconnectable Star Coupler so that bidirectional communication, which can add confidentiality to the broadcasting bus and doubles communication capacity, is available. A new configuration of Coupled Star Network using passive Interconnectable Star Couplers and optical amplifiers is proposed. This network has two separated bidirectional communication channel which can be applied for so-called Multimedia LAN. As a result of comparison between Cascade Star topology and Stratifide Star topology, it is shown that the latter topology is superior to former topology, from the view point of signal degeneration and maximum round trip delay time. Cascade Star topology, however, is superior to Stratified Star topology from the angle of total fiber length. Accordingly, optimized network topology is thought to be mixed topology of these.