Resistive field plates achieve high breakdown voltage when merged with an ordinary metal field plate. The optimum design of the structure was studied by a two dimensional device model and experiments. It was found that a properly combined structure realizes not only a higher breakdown voltage for shallow junctions but also a higher area efficiency, compared with conventional guard ring technologies.

This letter presents an approximation method for calculating the blocking probabilities of loss systems with batch input. Using this method, a batch stream can be treated in the same manner as an overflow stream because both streams can be characterized by common parameters: the offered load and peakedness.

A demountable optical connector for a single-mode fiber-ribbon has been developed, using a multi-V-groove molding method. The connecting loss due to the errors in ferrule fabrication has been analyzed in detail. The fabricated 5-fiber connector exhibited an average connecting loss of 0.5 dB.

It was shown experimentally that the noise, which is estimated to be due to mode hopping among modes composed of a laser mirror and an external reflection point, was suppressed in distributed-Bragg-reflector(DBR) single mode lasers biased above a certain level. This tendency agree with our theoretical prediction.

In the design of systolic algorithms, optimizing the speed of I/O operations and computations is an important problem. In this paper we develop a time-optimum systolic simulation algorithm which converts any two-way cellular algorithm into systolic one, where the former separates the computations from its I/O operations, on the other hand, the latter overlaps them. It is shown that, for any kn linear-time two-way cellular automaton M, there exists a systolic array which can simulate M in (k+2)n+O(1) steps. The parallel steps required for the simulation are optimum.

A new traffic forecasting method using state space representation is proposed. By means of a state space model, the forecasting value is sequentially calculated by applying the Kalman filter. However the true traffic structure is not easy to grasp as changes in traffic are largely due to social activities. In addition, experience has shown that economic trends in society also have an influence on traffic. For this reason, the traffic structure becomes too complex to describe changes in traffic by using a single state space model. In this paper a multiple state space model is proposed. The multiple state space models is composed of several state space models calls sub-models. This model is more easily adaptable to change in the traffic structure than a single state space model. The Bayesian forecasting value is given by the weighted summation for each sub-model forecasting value. The Bayesian posterior probability, which is calculated from the likelihood, is used as the weight of the sub-model. A good fitting sub-model posterior probability increases as the number of observations increases. In this paper the initial state and noise variances of each sub-model are estimated by numerical maximization of the likelihood. Examples of how this method may be applied to monthly telephone revenue data and trunk group load data are given, demonstrating the possibility of adapting exceptional data and structural changes in traffic. Parameter estimation using a multiple state space model is also shown.

The oscillatory-nonoscillatory transition is induced in blocking oscillator by controlling the source bias voltage. The oscillation peried shows logarhythmic divergence near the critical bias where the fluctuation of it also becomes remarkable. Its standard deviation is expressed by power law with critical index of 0.33.

We have presented a formulation of convective diffusion equations based on the method of sub-regions and the boundary element method, especially in three dimensions, in order to analyze very long and narrow convection-diffusion fields which we frequently encounter. In this formulation, we have introduced mixed-type boundary elements, which have more advantages in boundary element computation, and the weighted conservation law of mass in each sub-boundary model so as to be conservative in itself. In addition we have proposed a coupling technique at interface boundaries in order to equalize the number of equations to that of interface unknowns. As a result, it is found that the present method will give more accurate and stable solutions and also have more merits even on the basis of the method of sub-regions, and so is suitable and applicable particularly to a three-dimensional problem. The present formulation for mixed elements is also usable not only to convective diffusion problems, but also to Laplace and Helmholtz-type equations in steady and unsteady state.

We present here a new algorithm for checking equivalence of finite-turn dpda's. It is very direct and much simpler than the previous one by Valiant or by Beeri.

A temperature-tuned Fabry-Perot resonator fabricated by RF-sputtering Ag thin-film on both output end faces of a short piece of quasi-single mode fibers is presented. In order to analyze the modal properties of optical fibers composing a fiber-optic resonator, some unknown integers concerning with Fabry-Perot resonance modes are determined for a nominally circular-core fiber.

1.55µm GaInAsP/InP BH lasers with multiple p-n current blocking layer were fabricated entirely by low-pressure OMVPE. The undercut portions were filled in a manner similar to mass transport process. The laser shows a threshold current of 48 mA at 300 K under CW operation.

A new waterproof optical fiber cable is proposed which used absorbent polymer yarn instead of the conventional jelly-compound as a filling material. This yarn limits the length of water penetration into cable less than 10 cm that is short enough to guarantee long-term reliability. And it is found that the cable has superior characteristics for dynamic bend-tension test.

This paper describes experiments for evaluating the speech quality of packetized voice communications. The experimental conversations are performed through experimental packetized voice communications equipment, which provides packet transmission delays and packet losses. Three experimental results are apparent: packet losses degrade speech quality much more than packet transmission delays in PCM, packet delays affect speech quality more than packet losses in ADM, and inserting the last received codes for lost packets reduces the degradation of speech quality in PCM.

This paper describes a design method for a 446 Mb/s optical repeater with four kinds of monolithic integrated circuits. It is mainly concerned with the circuit division, optical receiver and timing amplifier design methods. Repeater circuit is divided based on maximum allowable gain and power consumption. Allowable gain determined by IC package crosstalk characteristics at 400 MHz transmission band is presented here. Allowable maximum power consumption of 1 W/chip is also determined. Effective amplification bandwidth adjustment methods for the optical receiver and the d.c. drift control (DCFB) system design method, specific in the d.c. coupled monolithic integrated optical receiver, are described. Finally, a useful design method for small-phase-shift timing amplifier is proposed. This is a circuit parameter optimizing method which monitors operating point dependence of small signal amplification bandwidth. The 446 Mbit/s submerged optical repeater has successfully been developed using these design methods.

In the Morita and Higuti (M-H) attenuation prediction method, the spatial correlation function of specific attenuation for conditional time, ρΓ, is, approximated by ρ, the spatial correlation function of rainfall rate. The accuracy of this approximation is examined in this paper. An equation relating ρΓ to ρ is theoretically derived assuming a bi-lognormal distribution of rainfall rates at two points. Theoretical examination shows that the approximation ρΓ=ρ does not necessarily hold accurately in the frequency range 10 to 100 GHz, except for frequencies near 30 GHz. This is also confirmed experimentally by measurements from a rain gauge network. A modified M-H method in which attenuation is predicted using ρΓ is proposed, and an extensive numerical comparison between both methods is made. It is shown that the approximation ρΓ=ρ can limit the applicability of the M-H method in terms of rain climatic condition, frequency, and path length, and that the modified M-H method has the advantage of improved applicability, compared with the M-H method.

To utilize the common-key encryption foe the message protection in a communication network, it is desired to settle the problem of how to distribute the common keys. This paper discusses a sort of schemes (the direct schemes, we call) that smartly provide different keys in different communications. Such a property has not attained via the basic scheme for the public key distribution systems proposed by Diffie and Hellman. This paper shows that the recently introduced five direct schemes are classified into three sets (called sequences) of infinite schemes, and points out that there are some tight relations among the sequences. And it is clarified which is the best in the three sequences by a systematic evaluation of the complexities for the normal update and for the deliberate forgery of the shared common keys.

The letter deals with theoretical analysis on Lamb waves propagating in a composite membrane of ZnO and SiO2. Calculation was done for the lowest pseudo symmetric (S0) and antisymmetric (A0) modes. It was shown that, for example, when the thickness of each film is of the order of 0.1 wavelengths or less, S0 mode has a large electromechanical coupling coefficient of about 6% and a velocity of more than 5,000 m/s. A0 mode, as may be expected, has a quite small velocity of less than 1,500 m/s with an electromechanical coupling coefficient of about 3%. The result suggests that it may be possible to realise various monolithic acoustic wave devices based on Lamb waves propagating in the ZnO/SiO2 composite membrane.

We show that if a signal (random or deterministic) is sampled nonuniformly (with Poisson or uniform distribution) and then low pass filtered, the output signal is equal to the input signal plus an additive white noise uncorrelated to the input signal. Hence, the power spectrum of random nonuniform samples and SNR is derived. Similar results can be derived for uniform sampling with some missing samples or with time jitter. We finally, show that the original signal can be recovered exactly (in the mean squared error sense) from random nonuniform samples by an iterative method provided that the average sampling rate is greater than the Nyquist rate.

A new boundary integral formulation is presented in order to solve a general Laplace-Poisson's equation, which is one of the basic equations of semiconductor devices. As this formulation is based on Green's second identity or Gauss' divergence theorem, no conventional volume integral is needed, regardless of arbitrary distributions of space charge. The potentials and electric field intensities at interface nodes put between a Laplace and a Poisson domain are analytically calculated, because interface nodes are treated as same as internal points. It is effective and powerful to device analysis of such a junction-gate field effect transistor with interfaces movable according to operation bias conditions. On the basis of simple numerical experiments, the present method is applied to a simplified device models. It is shown that device analysis can be easily obtained for a more small discretized model. In consequence, numerical results also demonstrate the effectiveness of this approach.

Four sets of on-board experiments regarding the measurements of 1.5 GHz multipath fading on propagation paths with low elevation angles and evaluation of fading reduction effects using the polarization shaping method were carried out. This paper describes the outline of the experiments, results obtained, and the comparison with theoretical values of the model presented by the authors. Results indicate that fading depth hardly depends on the wave height when the wave height is higher than 1 m, and that the polarization shaping method is effective for both up-link and down-link fadings for sea conditions with a wave height of up to at least 3 m.