The flip-chip structure for millimeter-wave MMICs has been investigated to obtain high performance and high reliability. In our approach, an air gap between the MMIC and the alumina substrate was determined so as not to change electrical characteristics from those of the unflipped MMIC. We calculated the proximity effect between the MMIC and the substrate by using 3D-electromagnetic simulator, and found that the air gap should be controlled to be greater than 20 µm. Since the discontinuity of transmission lines at bump interconnects is not negligible above 60 GHz, we constructed the LCR-equivalent circuit for the bump interconnect and confirmed its validity by comparing measurement with calculation. Based on these investigations, the 60- and 76-GHz-band CPW three-stage low noise amplifiers were successfully mounted on the alumina substrate using a thermal compression bonding process. The gain of the flipped 60- and 76-GHz-band MMICs are greater than 18 dB at around 60 GHz and 17 dB at around 76 GHz, respectively. The noise figures are 3.6 dB and 3.9 dB, respectively. The gain and noise performances showed little degradation compared to those of the unflipped MMICs when appropriate bonding conditions are given. We confirmed that the flip-chip structure has high reliability under a thermal cycle test. From these results, flip-chip technology is promising for millimeter-wave applications.

The glottal closure instants (GCIs) obtained from the wavelet analysis of speech signal are investigated in comparison with those obtained from the EGG signal. Experimental results have shown that about 96% of GCIs with wavelet transformed speech signal is located within 0.5 ms with respect to the GCIs with EGG signal.

In this paper, we consider sufficient conditions for ruling out some useless iteration steps in a class of soft-decision iterative decoding algorithms for binary block codes used over the AWGN channel using BPSK signaling. Sufficient conditions for ruling out the next single decoding step, called ruling-out conditions and those for ruling out all the subsequent iteration steps, called early termination conditions, are formulated in a unified way without degradation of error performance. These conditions are shown to be a type of integer programming problems. Several techniques for reducing such an integer programming problem to a set of subprograms with smaller computational complexities are presented. As an example, an early termination condition for Chase-type decoding algorithm is presented. Simulation results for the (64, 42, 8) Reed-Muller code and (64, 45, 8) extended BCH code show that the early termination condition combined with a ruling-out condition proposed previously is considerably effective in reducing the number of test error patterns, especially as the total number of test error patterns concerned grows.

The growth behavior of copper particle on crystalline and amorphous silicon surfaces has been investigated. The study reveals that the growth behavior of copper particle depends on the substrate condition. When samples are intentionally contaminated in ultrapure water, both crystalline and amorphous silicon surfaces show no difference in their contamination levels. However, copper particles were not observed on an amorphous silicon surface except dipping in dilute CuCl2 solution. The copper concentration on an amorphous silicon surface after dipping in a 0.5% HF solution is similar to the level after contaminating in ultrapure water. The copper contamination level on a crystalline silicon surface, except from CuCl2 solution, decreased two orders of magnitude as compared with ultrapure water. The copper impurity level on crystalline silicon surface was reduced by two orders by cleaning in a sulfuric acid-hydrogen peroxide mixture. The sulfuric acid-hydrogen peroxide mixture cleaning was not effective on an amorphous silicon surface. When native oxide pre-existed on an amorphous silicon surface before contamination, however, the sulfuric acid-hydrogen peroxide mixture cleaning was effective for removing copper impurity. Our results suggest that copper contamination on an amorphous silicon surface have the characteristics of bonding directly with silicon and/or existing in the native oxide, in contrast with the situation on crystalline silicon surface. After contamination with 1000 ppm copper in CuF2 solution, the etch rate of an amorphous silicon film in a 0.5% HF solution was approximately one order of magnitude faster than that of crystalline silicon. This is attributed to the difference in crystalline structure between crystalline silicon and amorphous silicon.

In this paper, we present simulation algorithms among enhanced mesh models. The enhanced mesh models here include reconfigurable mesh and mesh with multiple broadcasting. A reconfigurable mesh (RM) is a processor array that consists of processors arranged to a 2-dimensional grid with a reconfigurable bus system. The bus system can be used to dynamically obtain various interconnection patterns among the processors during the execution of programs. A horizontal-vertical RM (HV-RM) is obtained from the general RM model, by restricting the network topology it can take to the ones in which each bus segment must be along row or column. A mesh with multiple broadcasting (MWMB) is an enhanced mesh, which has additional broadcasting buses endowed to every row and column. We present two algorithms:1) an algorithm that simulates a HV-RM of size nn time-optimally in θ(n) time on a MWMB of size nn, and 2) an algorithm that simulates a RM of size nn in θ(log2 n) time on a HV-RM of size nn. Both algorithms use a constant number of storage in each processor. Furthermore, we show that a RM of size nn can be simulated in θ((n/m)2 log n log m) time on a HV-RM of size mm, in θ ((n/m)2 m log n log m) time on a MWMB of size mm (m < n). These simulations use θ((n/m)2) storage in each processor, which is optimal.

This paper presents a performance evaluation of NAK-based reliable multicast communication protocols operating in an environment where end-to-end delay are heterogeneous. In the case of heterogeneous delay, performance of a timer-based retransmission control scheme may become worse. We show that a counter-based retransmission control scheme works well in the case of heterogeneous transmission delay. We also compare two NAK-based protocols and show that a NAK-multicasting protocol outperforms a NAK-unicasting protocol from the viewpoint of scalability even when delays are heterogeneous.

Dynamical theory of cellular automata on groups is developed. Main results are non-Euclidean extensions of Sato and Honda's results on the dynamics of Euclidean cellular automata. The notion of the period of a configuration is redefined in a more group theoretical way. The notion of a co-finite configuration substitutes the notion of a periodic configuration, where the new term is given to it to reflect and emphasize the importance of finiteness involved. With these extended or substituted notions, the relations among period preservablity, injectivity, and Poisson stability of parallel maps are established. Residually finite groups are shown to give a nice topological property that co-finite configurations are dense in the configuration space.

In pattern recognition using neural networks, it is very difficult for researchers or users to design optimal neural network architecture for a specific task. It is possible for any kinds of neural network architectures to obtain a certain measure of recognition ratio. It is, however, difficult to get an optimal neural network architecture for a specific task analytically in the recognition ratio and effectiveness of training. In this paper, an evolutional method of training and designing feedforward neural networks is proposed. In the proposed method, a neural network is defined as one individual and neural networks whose architectures are same as one species. These networks are evaluated by normalized M. S. E. (Mean Square Error) which presents a performance of a network for training patterns. Then, their architectures evolve according to an evolution rule proposed here. Architectures of neural networks, in other words, species, are evaluated by another measurement of criteria compared with the criteria of individuals. The criteria assess the most superior individual in the species and the speed of evolution of the species. The species are increased or decreased in population size according to the criteria. The evolution rule generates a little bit different architectures of neural network from superior species. The proposed method, therefore, can generate variety of architectures of neural networks. The designing and training neural networks which performs simple 3 3 and 4 4 pixels which include vertical, horizontal and oblique lines classifications and Handwritten KATAKANA recognitions are presented. The efficiency of proposed method is also discussed.

This paper describes a new method for detecting the gaze position of a user on a monitor from monocular images. In order to detect the gaze position, we extract facial features (both eyes, nostrils and lip corners) automatically in 2D camera images and estimate the 3D depth information and the initial 3D positions of those features by recursive estimation algorithm in starting images. Then, when a user moves his/her head in order to gaze at one position on a monitor, the moved 3D positions of those features can be estimated from 3D motion estimation by Extended Kalman Filter (EKF) and affine transform. Finally, the gaze position on a monitor is calculated from the normal vector of the plane determined by those moved 3D positions of features. Especially, in order to obtain the exact 3D depth and positions of initial feature points, we unify three coordinate systems (face, monitor and camera coordinate system) based on perspective transformation. As experimental results, the 3D depth and the position estimation error of initial feature points, which is the RMS error between the estimated initial 3D feature positions and the real positions (measured by 3D position tracker sensor) is about 1.28 cm (0.75 cm in X axis, 0.85 cm in Y axis, 0.6 cm in Z axis) and the 3D motion estimation errors of feature points by Extended Kalman Filter (EKF) are about 3.6 degrees and 1.4 cm in rotation and translation, respectively. From that, we can obtain the gaze position on a monitor (17 inches) and the gaze position accuracy between the calculated positions and the real ones is about 2.1 inches of RMS error.

We proposed a soft-decision decoding algorithm for cyclic codes based on energy minimization principle. This letter presents the algorithm which improves decoding performance and decoding complexity of the previous method by giving more initial positions and introducing a new criterion for terminating the decoding procedure. Computer simulation results show that both the decoded block error rate and the decoding complexity decrease by this method more than by the previous method.

In this paper, we study the link capacity assignment problem in packet-switched networks (CA problem) focusing on the case where link cost function is a piecewise linear concave function. This type of cost function arises in many communication network design problems such as those arising from developments in communication transmission technologies. It is already known that the method of link set assignment is applicable for solving the CA problem with piecewise linear convex cost function. That is, each link in the network is assigned to one of a group of specific sets, and checked for link set contradiction. By extending the method of link set assignment to the case of piecewise linear concave cost function, an important characteristic of the optimal solution of the CA problem is derived. Based on this characteristic, the non-differentiable link cost function can be treated as a differentiable function, and a heuristic algorithm derived from the Lagrange multiplier method is then proposed. Although it is difficult to determine the global optimum of the CA problem due to its non-convexity, it is shown by numerical results that the solution obtained from the proposed algorithm is very close to the global optimum. Moreover, the computation time is linearly dependent on the number of links in the problem. These performances show that the proposed algorithm is very efficient in solving the CA problem, even in the case of large-scale networks.

For the realization of low-voltage full-color FEDs, requirements for phosphor for the FED are proposed. Especially, the influence of released gases or substances from phosphors on the field emission within the FED was made clear. It was clarified that the analysis of F-N plots of the V-I curve of field emission characteristics was helpful to know the interaction of field emission and phosphors. In the experiment, we first obtained the depth from the phosphor surface of the low voltage electron excitation in case of ZnGa2O4, where the region available for cathodoluminescence at the anode voltage of 400 V is about 63 nm deep from the surface. The characteristic of the 12.4 cm-320(trio)240 pixels low-voltage full-color FED is reported. The luminance of 154 cd/m2 was attained at the anode voltage of 400 V and the duty factor of 1/241. Supported by the high potential of the FED as a flat panel, each problem shall be steadily solved to secure the firm stand as a new full color flat display in new applications.

We propose a traffic measurement system which uses trap and polling methods. To obtain its performance we consider a queueing model with a single server and evaluate a packet delay. In our multi-cast traffic, packets are modeled as a batch with a batch size distribution {gk}. The batch arrival process is observed as two processes on the basis of batch size. For a batch whose size is more than or equal to a threshold L, the batch will be trapped by our traffic measurement system (in queueing model, it will enter a queue immediately after its arrival). For a batch whose size is less than L, it will be observed at a polling cycle T (in queueing model, it will be temporarily stored in a buffer and all these small batches will be cyclically noticed with a cycle T). We analyze this queueing model by a diffusion approximation and compare the packet delay observed by our traffic measurement system with the L=1 original batch arrival model. Evaluating the results of the diffusion approximation, we illustrate that our traffic measurement system has functions not only to give an accurate estimation of the mean waiting time but also reduce the number of measurements by choosing appropriate parameters L and T.

This paper describes numerical analyses of resistive mixer operation, followed by measured performances of a V-band (50 - 75 GHz) monolithic InP HEMT resistive mixer operable with a very low LO power. Our model assumes that the channel conductance of the InP HEMT can be described by three linear functions according to the applied gate voltage. The calculated results obtained with the model have shown that the LO power level required for mixer operation is determined by the gate bias voltage and that a device with abrupt conductance shifts is suited to low LO power operation for a resistive mixer. It is also shown that conversion loss saturation of a resistive mixer is caused by its channel conductance saturation. A V-band monolithic resistive mixer has been designed and fabricated using Coplanar Waveguides (CPW) and a 0.15 mm InP HEMT with abrupt channel shifts. Good agreement between measured and simulated conversion losses are obtained. A minimum conversion loss of 8.4 dB is achieved at the 55 GHz RF-frequency and the -2 dBm LO power. It also exhibits an excellent IF output linearity to allow the 1 dB compression RF input level to be comparable with LO power, indicating good intermodulation performance. It is demonstrated that the proposed simple model of the channel conductance can easily calculate conversion characteristics of a resistive mixer with high accuracy.

We investigated the relationship between the film characteristics and the sputtering rate of the MgO protecting layer in AC-PDP. As possible elements for determining the sputtering rate, we considered the density, orientation, and surface morphology. With respect to the orientation, we found that the sputtering rate increased for the sequence of (200) < (220) < (111). However, we noticed that orientation and surface structure are not really decisive factors affecting the sputtering rate; the density of the film is most important.

The relationships between lattice orientation of the electron-beam evaporated MgO layer used as protecting layer for ac plasma displays (ac-PDPs) and the discharge characteristics of color ac-PDPs were investigated by the measurements of ion-induced secondary electron emission. It is proved that values of γi for MgO are large in the order of (220) orientation, (200) orientation, and (111) orientation, that is, γi(220) > γi(200) > γi(111). The values of φ for different lattice orientation are obtained by the measurements of thermionic emission and photo emission. The aging measurements for testing panels with the different lattice orientation of MgO layer revealed that performance of those panels are excellent in the order of (220), (200), and (111). In particular, luminance and luminous efficiency become larger in the order of (220), (200), and (111). It is pointed out that the degree of longevity, sustaining voltage, and memory margin for ac-PDPs with protecting materials as MgO are estimated by the measurements of γi.

Experiment is reported of enhanced backscattering of light in binary and ternary suspensions of rutile and/or alumina particles. With a conventional CCD camera system for observing the phenomena, the angular line shape and the enhancement factor were agreed with the theoretically predicted curve and value. Observation of the angular distribution scattered at the backscattered direction supported the hypothesis proposed by Pine et al. , in which the transport mean free path of the polydisperse mixture can be expressed in terms of summing its reciprocal values weighted over the particle sizes.

In this paper, we describe an experimental evaluation of visual fatigue in a binocular disparity type 3-D display system. To evaluate this fatigue, we use a subjective assessment method and focus on mismatching between convergence and accommodation, which is a major weakness of binocular disparity 3-D displays. For this subjective assessment, we use a newly-developed binocular disparity 3-D display system with a compensation function for accommodation. Because this equipment only allowed us to compare the terms of the mismatching itself, the evaluation is more accurate than similar previous works.

Optical Code Division Multiple Access (OCD- MA) has been emerging as an attractive scheme in fiber optic communication systems as well as in space communication systems in past few years. In OCDMA systems, M-ary Pulse Position Modulation (PPM), has been regarded as an efficient signalling format which has the capability to reduce the channel interference caused by the other users and also to increase the number of simultaneous users. We apply error control coding to improve the system performance of pulse position modulated OCDMA (PPM/OCDMA) systems and this paper investigates the performance of M-ary PPM/OCDMA systems with M-ary convolutional coding. Dual-k code is used as the M-ary convolutional code and Optical Orthogonal Codes with the maximum cross correlation value of 1 and 2 are employed as the signature sequences. We derive an expression for the bit error probability of the new system and show that combining M-ary convolutional coding and M-ary PPM results in an improved error performance. Also it is shown that the number of simultaneous users can be significantly increased with the proposed system compared to the uncoded PPM/OCDMA system with the same bit error probability and with the same information bit rate. We also analyze the system with binary convolutional coding and a comparison with the proposed system is given.

We analyze the extended stochastic complexity (ESC) which has been proposed by K. Yamanishi. The ESC can be applied to learning algorithms for on-line prediction and batch-learning settings. Yamanishi derived the upper bound of ESC satisfying uniformly for all data sequences and that of the asymptotic expectation of ESC. However, Yamanishi concentrates mainly on the worst case performance and the lower bound has not been derived. In this paper, we show some interesting properties of ESC which are similar to Bayesian statistics: the Bayes rule and the asymptotic normality. We then derive the asymptotic formula of ESC in the meaning of almost sure and mean convergence within an error of o(1) using these properties.