In this paper partial response signalling and trellis coded modulation are considered together to improve bandwidth efficiency and error performance for M-QAM and denoted as Modified/Quadrature Partial Response-Trellis Coded Modulation (M/QPR-TCM) and two new non-catastrophic schemes M/6QPR-TCM and M/9QPR-TCM are introduced for 4QAM. In colored noise with correlation coefficient less than zero, the proposed schemes perform better than in AWGN case. Another interesting result is that when the combined system is used on a Rician fading channel, the bit error probability upper bounds of the proposed systems are better than their counterparts the 4QAM-TCM systems with 2 and 4 states, respectively, for SNR values greater than a threshold, which have the best error performance in the literature.

Entire systems on a chip (SOCs) embodying a processor, memory, and system-specific peripheral hardware are now an everyday reality. The current generation of SOC designers are driven more than ever by the need to lower chip cost, while at the same time being faced with demands to get designs to market more quickly. It was to support this new community of designers that we developed Satsuki-an integrated processor synthesis and compiler generation system. By allowing the designer to tune the processor design to the bitwidth and performance required by the application, minimum cost designs are achieved. Using synthesis to implement the processor in the same technology as the rest of the chip, allows for global chip optimization from the perspective of the system as a whole and assures design portability. The integral compiler generator, driven by the same parameters used for processor synthesis, promotes high-level expression of application algorithms while at the same time isolating the application software from the processor implementation. Synthesis experiments incorporating a 0.8 micron CMOS gate array have produced designs ranging from a 45 MHz, 1,500 gate, 8-bit processor with a 4-word register file to a 31 MHz, 9,800 gate, 32-bit processor with a 16-word register file.

Gray scale images are represented by recurrent neural subnetworks which together with a competition layer create an associative memory. The single recurrent subnetwork Ni implements a stochastic nonlinear fractal operator Fi, constructed for the given image fi. We show that under realstic assumptions F has a unique attractor which is located in the vicinity of the original image. Therefore one subnetwork represents one original image. The associative recall is implemented in two stages. Firstly, the competition layer finds the most invariant subnetwork for the given input noisy image g. Next, the selected recurrent subnetwork in few (5-10) global iterations produces high quality approximation of the original image. The degree of invariance for the subnetwork Ni on the inprt g is measured by a norm ||g-Fi(g)||. We have experimentally verified that associative recall for images of natural scenes with pixel values in [0, 255] is successful even when Gaussian noise has the standard deviation σ as large as 500. Moreover, the norm, computed only on 10% of pixels chosen randomly from images still successfuly recalls a close approximation of original image. Comparing to Amari-Hopfield associative memory, our solution has no spurious states, is less sensitive to noise, and its network complexity is significantly lower. However, for each new stored image a new subnetwork must be added.

The problem of electromagnetic scattering by capacitive discontinuities located in rectangular waveguides, in particular when dealing with discontinuous conductors of finite thickness, is analyzed using the modified residue-calculus method, and form of the equation suitable for a numerical calculation is derived. The incident wave is taken to be the dominant mode, and reflection and transmission properties of an asymmetric capacitive iris are discussed. After the modal representation of the filed, the modal matching is apply to satisfy the boundary conditions at the discontinuity. And using the modified residue-calculus method, simultaneous infinite equations, which are concerned with the scattered mode coefficients, are derived. Then they are approximated at the thick diaphragm. The solutions obtained take on the form of an infinite product, and a numerical solution based on the method of successive approximations is presented as a technique for concretely determining the reflection coefficients. As confirmation, experiments are also carried out in the X-band and close agreement is shown between the calculated and experimental values.

This paper presents the scattering characteristics of a TE electromagnetic plane wave by metallic strip gratings on an optically plasma-induced silicon slab at millimeter wave frequencies. The characteristics were analyzed by using the spectral domain Galerkin method and estimated numerically. We examined to control the resonance anomaly by changing the optically induced plasma density, and the metallic strip grating structures were fabricated on highly resistive silicon. The optical control characteristics of the reflection, and the forward scattering pattern by the grating structures, were measured at Q band and are discussed briefly with theory.

In order to apply formal design verification, it is necessary to describe formally and correctly the specification of the circuit under verification. Especially when we apply conventional OBDD-based logic comparison method for verifying combinational circuits, another correct" logic circuits or Boolean formulae must be given as the specification. It is desired to develop an efficient automatic design verification method which interprets specification that can be described easier. This paper provides a new verification method which is useful for combinational circuits such as arithmetic circuits. The proposed method efficiently verifies whether a designed circuit satisfies a specification given by recurrence equations. This enables us to describe easily an error-free specification for arithmetic circuits. To perform verification efficiently using an ordinary OBDD package, an efficient truth-value rotation algorithm is developed. The truthvalue rotation algorithm efficiently generates an OBDD representing f(x + 1 (mod 2n)) from a given OBDD representing f(x). By experiments on SPARC station 10 model 51, it takes 180 secs to generate an OBDD for designed circuit of 23-bit square function, and additional 60 secs is sufficient to finish verifying that it satisfies the specification given by recurrence equations.

Synthesis tools are now extensively used in the VLSI circuit design process. They allow a much higher design productivity, but the designer often does not directly control the circuit structure. Thus, when circuits are dedicated to dependable applications, designers have difficulties in implementing manually the devices needed to obtain fault detection or tolerance capabilities. The ASYL-SdF System has been developed over the last few years in order to avoid this break in the design flow, and to facilitate the designer's work when dependability is targeted. This paper gives an overview of the resulting tool, its synthesis flow for fault detection and fault tolerance in Finite State Machines, its limitations and the current developments. Actual circuit implementation results are given in terms of area overheads, expected reliability and experimental fault detection coverage.

Moving-picture transmission through narrow band and high bit error rate communication channels, such as a mobile communication channel, requires improved compression rate and enhanced error resilience. Variable-length codes are one of the essential techniques of compressing digital video information. This technique is used in various video coding schemes although a bit error in the channel impairs the synchronization of variable-length codewords, resulting in propagation of the error. With a hybrid video coding method in particular, which combines motion-compensation and transform coding, once an error is detected in the coded data, subsequent data cannot be decoded. Consequently, even an error-free portion of any data received must be discarded. To minimize the influence of an error in a channel on coded video data, this paper proposes a new video coding syntax which makes the best use of the self synchronizing characteristic of variable-length Huffman codes. Owing to the Huffman code's characteristic, the proposed coding syntax enables a decoder to decode the data portion that cannot be decoded, due to an error, by the conventional syntax without adding any redundancy. Computer simulation has verified the effectiveness of this proposed syntax in video coding with a very low bitrate and erroneous communication channel.

3D model-based coding methods that need 3D reconstruction techniques are proposed for next-generation image coding methods. A method is presented that reconstructs 3D shapes of dynamic objects from image sequences captured using two cameras, thus avoiding the stereo correspondence problem. A coaxial camera system consisting of one moving and one static camera was developed. The optical axes of both cameras are precisely adjusted and have the same orientation using an optical system with true and half mirrors. The moving camera is moved along a straight horizontal line. This method can reconstruct 3D shapes of static objects as well as dynamic objects using motion vectors calculated from the moving camera images and revised using the static camera image. The method was tested successfully on real images by reconstructing a moving human shape.

The author once defined the Ω-matrix and showed that it played an important role for estimating the number of solutions of a resistive circuit containing active elements such as CCCS's. The Ω-matlix is a generalization of the wellknown P-matrix. This paper gives the necessary and sufficient conditions for the Ω-matrix.

This paper verifies the accuracy of PO as applied to the scattering of dipole waves by a finite size reflector which is composed of strips on a grounded dielectric slab. By using the closed form expressions of reflected waves from the surface, PO calculation can be conducted straightforwardly. The calculated results are compared with the experimental ones for vertical and horizontal dipoles over a circular reflector.

A method is presented for reconstructing the surface profile of a two dimensional rough surface boundary from the scattered far field data. The proposed inversion algorithm is based on the Kirchhoff approximation and in order to determine the surface profile, the numerical results illustrating the method are presented.

In order to improve the sensitivity of micromachined sensors applied with electrostatic fields and increase their actuated force of electrostatic micromachined actuators, "electrets," which are dielectrics carrying non equilibrium permanent space charges of polarization distribution, are very important. In this paper, positively corona charged silicon dioxide electrets, which are deposited by Plasma Chemical Vapor Deposition (PCVD) and thermally oxidized, are investigated. Physical studies will be described, in which the charge stability is correlated to Thermally Stimulated Current (TSC) measurements and to Electron Spin Resonance (ESR) analysis. Some intrinsic differences have been observed between materials. The electrets with superior long-term charge stability contain 10,000 times as much E' center (Si3 as the ones with inferior long-term charge stability. Finally, some investigations on the long-term charge storage mechanism of the positively charged silicon dioxide electret will be described.

An efficient algorithm is proposed for finding all solutions of piecewise-linear resistive circuits The algorithm is based on the idea of "contraction" of the solution domain using a sign test. The proposed algorithm is efficient because many large super-regions containing no solution are eliminated in early steps.

Deterministic nonlinear prediction is applied to both artificial and real time series data in order to investigate orbital-instabilities, short-term predictabilities and long-term unpredictabilities, which are important characteristics of deterministic chaos. As an example of artificial data, bimodal maps of chaotic neuron models are approximated by radial basis function networks, and the approximation abilities are evaluated by applying deterministic nonlinear prediction, estimating Lyapunov exponents and reconstructing bifurcation diagrams of chaotic neuron models. The functional approximation is also applied to squid giant axon response as an example of real data. Two metnods, the standard and smoothing interpolation, are adopted to construct radial basis function networks; while the former is the conventional method that reproduces data points strictly, the latter considers both faithfulness and smoothness of interpolation which is suitable under existence of noise. In order to take a balance between faithfulness and smoothness of interpolation, cross validation is applied to obtain an optimal one. As a result, it is confirmed that by the smoothing interpolation prediction performances are very high and estimated Lyapunov exponents are very similar to actual ones, even though in the case of periodic responses. Moreover, it is confirmed that reconstructed bifurcation diagrams are very similar to the original ones.

Conventional learning algorithms are considered to be a sort of estimation of the true recognition function from sample patterns. Such an estimation requires a good assumption on a prior distribution underlying behind learning data. On the other hand the human being sounds to be able to acquire a better result from an extremely small number of samples. This forces us to think that the human being might use a suitable prior (called presupposition here), which is an essential key to make recognition machines highly flexible. In the present paper we propose a framework for guessing the learner's presupposition used in his learning process based on his learning result. First it is pointed out that such a guess requires to assume what kind of estimation method the learner uses and that the problem of guessing the presupposition becomes in general ill-defined. With these in mind, the framework is given under the assumption that the learner utilizes the Bayesian estimation method, and a method how to determine the presupposition is demonstrated under two examples of constraints to both of a family of presuppositions and a set of recognition functions. Finally a simple example of learning with a presupposition is demonstrated to show that the guessed presupposition guarantees a better fitting to the samples and prevents a learning machine from falling into over learning.

The capability of generalization is the most desirable property of a learning system. It is well known that to achieve good generalization, the complexity of the system should match the intrinsic complexity of the problem to be learned. In this work, introduction of fractal connection structure in nonlinear learning systems like multilayer perceptrons as a means of improving its generalization capability in classification problems has been investigated via simulation on sonar data set in underwater target classification problem. It has been found that fractally connected net has better generalization capability compared to the fully connected net and a randomly connected net of same average connectivity for proper choice of fractal dimension which controlls the average connectivity of the net.

Most research on the estimation of direction of arrival (DOA) has been performed based on the assumption that the signal sources are point sources. In some real surroundings, signal source localization can more adequately be accomplished with distributed source models. When the signal sources are distributed over an area, we cannot directly use well-known DOA estimation methods, because these methods are established based on the point source assumption. In this paper, we propose a 3-dimensional distributed signal source model, in which a source is represented by two parameters, the center angle and degree of dispersion. Then, we address the estimation of the elevation and azimuth angles of distributed sources based on the parametric distributed source modeling in the 3-dimensional space.

There are many kinds of transmission lines such as uniform, nonuniform and nonlinear ones terminated by linear and/or nonlinear subnetworks. The nonuniform transmission lines are crucial in integrated circuits and printed circuit boards, because these circuits have complex geometries and layout between the multi layers, and most of the transmission lines possess nonuniform characteristics. On the other hand, the nonlinear transmission line have been focused in the fields of communication and instrumentation. Here, we present a new numerical method for analyzing nonuniform and nonlinear transmission lines with linear and/or nonlinear terminations. The waveforms at any points along the lines are described by the Fourier expansions. The partial differential equations representing the circuit are transformed into a set of ordinary differential equations at each frequency component, where for nonlinear transmission line, the perturbation technique is applied. The method is efficiently applied to weakly nonlinear transmission line. The nonuniform transmission lines terminated by a nonlinear subnetwork are analyzed by hybrid frequency-domain method. The stability for stiff circuit is improved by introducing compensation element. The efficiency of our method is illustrated by some examples.

Several two-dimensional largest common subpatterns (LCP) between pictures are defined and their computing methods are proposed. The time and space complexities of the computing methods are O(IJMN) to obtain the size of LCPs between a picture with IJ pixels and a picture with MN pixels. These LCPs can be used as similarity measures between pictures and can be applied to texture recognition and classification.