A generalized conservation law for mixed preemptive and nonpreemptive discipline is presented. In this discipline there are N job classes which are assigned preemptive resume priority and each job class split into Nj(j0, , N) job subclasses which are assigned head-of-the-line priority. The discipline is used, e.g., in communication control in a switching system. We give a proof for the conservation law by an intuitively comprehensive direct method, which is also applied for the Kleinrock's conservation law to make his proof elaborated. An alternative proof for Takahashi's conservation law on preemptive-resume M GI/G1 G2/1 systems is also presented to clarify the physical meaning in the connection with completion times.

The nonuniformity of signal constellation on coded phase/frequency modulation is investigated to obtain further improvement on the error performance over conventional uniformly spaced coded phase/frequency modulation. The analytical method for evaluating the Euclidean distance of nonuniform FSK/PSK signal is presented. In particular, the joint design of m/(m1) systematic convolutional encoders and nonuniform 2FSK/2mPSK are considered, and especially for 2FSK/4PSK and 2FSK/8PSK the behavior of the free Euclidean distance is evaluated by varying the modulation index, the rotation angles which determine the nonuniform signal constellation, and the mapping rule by set partitioning. The results show that by designing compared with the conventional coded uniform phase/frequency modulation may be obtained without increasing average or peak power or changing the bandwidth of the transmitted signals. The maximum value of the free Euclidean distance of codes can be obtained when the PSK signal constellation of one carrier frequency is the same as those of the other carrier frequency, and the mapping rule used in coded uniform FSK/PSK system is not always good enough when the nonuniformity in the signal constellation is introduced.

Mass fusion splice without accurate fiber-core alignment for splicing single-mode fiber ribbon in subscriber optical cable is discussed. Using the surface tension effect in the fusion process positively and suppressing the fiber length variance among fibers in the ribbon, low splice loss of about 0.1 dB was achieved for fibers with small core eccentricity.

Taking into account the higher order dispersions of the complex propagation constant involving its loss term, the equation of a Gaussian pulse propagation is derived for a single-mode optical fiber, by which the pulse-peak intensity, the pulse-width, the pulse-delay time and the skewness can be readily estimated along the fiber length. Then, among various combinations of the dispersion profile between the loss and the phase terms including both normal and anomalous dispersions, two typical cases of the pulse-broadening and narrowing are numerically shown, in which possibility of the compression is especially discussed with the combination of the sharp anomalous absorption-band and the normal phase-dispersions.

A simple model for pattern recognition is proposed, which can be approximated by a multilayer network. An application to handwritten character recognition is simulated and characters were recognized with an accuracy about 90%. Using this model enables the system to be simpler.

This paper discusses the characteristics of consciousness structures as to the training of a turning lathe. For this purpose, a consciousness survey is forwarded to the trainers and the trainees. From the answered results, we make consciousness structures for each group, and then compare them. By comparing the characteristic of the trainees with that of trainers, it got clear that the trainees look more interested in worker's behavior such as cutting than the phenomenon on lathe such as heat or vibration, and so forth.

This letter presents the development of a high-speed 8-stage synchronous counter LSI which uses a 400-gate macrocell array. The LSI uses a sophisticated low-voltage-swing differential CML circuit technique and super-self-aligned Si-bipolar process technology (SST). The counter operates at up to 2 GHz with a chip power dissipation of 0.91 W.

Upon the close examination of some typical papers treating E-Plane Symmetrical T-junctions, we derive peculiar equivalent circuits of the T-junctions in terms of the direct sum representation of the three-port. We show that the circuit elements can be determined by use of the Weissfloch nodal-shift method and that the circuits give an unified description and some new physical meanings of Lewin's equivalent circuit. Allanson-Cooper-Cowling's one, and others. Then we also give an analytical derivation of the direct sum in terms of a system of integral equations rearranged in a hybrid matrix whose determinant is nearly equal to zero for a tee. This property of the hybrid matrix allows the number of independent parameters to be reduced by one in a good approximation. Further the stationary values of the variational equations equivalent to the integral equations straightforwardly give the circuits elements.

A method of applying the fast Fourier transform (FFT) algorithm to Filon quadrature is presented. The quadrature formula can be reduced to a linear combination of two vectors to each of which the ordinary FFT algorithm is ready to be applied. An expression for the truncation error of Filon quadrature is then derived to examine the numerical errors of the formula. The experimental results showed that error terms of up to O(h7), where h is the length of subintervals for Filon numerical integration, are required to explain the numerical errors observed in the spectral analysis of the response of a single-pole system to a periodic rectangular wave.

In this paper, we consider the performance of QPSK systems using complex transversal filters with additional decision-feedback (DF) taps, in the presence of colored noise and single continuous wave (CW) interference. Both one-sided and two-sided transversal filters with DF are considered. The general analytic expressions for the optimum tap weights and minimum mean square errors are obtained. We calculate the output signal-to-noise ratio of the filters and compare the theoretical results with the computer simulated results. The relations between the spectrum of the colored noise as well as CW interference and the corresponding frequency response of the DF filters are considered. We also consider the transient behavior of the adaptive DF filters. It is shown that the DF filters can suppress not only CW interference but also colored noise and that the two-sided transversal filter with DF has best performance in the presence of the colored noise and CW interference, while the one-sided transversal filter with DF is preferred when the white Gaussian noise and CW interference are included in the system. It is also shown that the frequency response of the DF filters is changed according to the spectrum distribution of the colored noise and CW interference in order to suppress them more effectively. The simulated results are in good agreement with the theoretical results.

Global stability of equilibrium states is investigated for a continuous-time model of neural networks and a discrete-time one. Three classes of globally stable networks are introduced. One class called weakly coupled networks is shown to be globally asymptotically stable, i.e. every trajectory eventually converges to a unique equilibrium point. The other two classes of which one is called gradient networks and the other is called type K monotone networks are guaranteed to be completely stable, i.e. any trajectory eventually converges to one of equilibrium states. These stability properties are preserved under introduction of any synaptic transmission delay. In addition monotone sensitivity is discussed for weakly coupled cooperative networks and type K monotone networks.

The concept of a contragredient mapping is introduced to analyze linear circuits. It enables us to apply Thévenis's theorem successively to each usual two-port with the simplest calculation.

A new photometric method, which can reconstruct 3-D space coordinate (i.e.Z-distribution) of an object from one image under a point light source illumination, is proposed. The object is continuous convex with the perfectly diffused surface and the known uniform reflectance. To get the Z-distribution by solving the illuminating equation basing on the inverse square law for illuminance, an iterative algorithm has been developed. The tangent plane of the brightest surface element is firstly determined. The objective Z-distribution is finally obtained by iterating processes of calculating the Z-distribution treating the gradient distribution as constant. At each iteration step, the Z-distribution for the next step is determined from the present Z-distribution and the calculated Z-distribution which satisfies the illuminating equation, and the gradient distribution for the next step is calculated geometrically from the determined Z-distribution basing on the continuity of the surface. The usefulness of this method has been demonstrated by computer simulations.

The mode analysis of an open-boundary Cerenkov laser is developed in the collective regime. The Cerenkov laser under consideration is composed of a magnetically-confined relativistic electron beam and a dielectric-loaded conducting plane. The electron beam and the dielectric are assumed to be arbitrary in thickness, with an arbitrary spacing allowed between them. For the Cerenkov laser specified above, the following results are obtained. First, an electromagnetic wave mode is coupled with an infinite number of space charge wave modes. In the general case, an electromagnetic wave mode is coupled collectively with space charge wave modes. On the other hand, in the special case where the coupling occurs near the Cerenkov threshold, an electromagnetic wave mode is coupled separately with each of space change wave modes. Second, the characteristics of the growing wave for a magnetically-confined beam are similar to those for an ion-neutralized beam, except that the magnitude of the spatial growth rate becomes somewhat smaller for the former than for the latter.

This paper proves several theorems on probabilistic cryptosystems. From these theorems it follows directly that a probabilistic cryptosystem proposed by the authors, whose security is based upon the (supposed) infeasibility of γth-Residuosity Problem, is polynomially secure. Techniques developed in the paper are of independent interest.

Frequency spectra of arc pulses in the gaseous phase arc of the surge are compared breaking Ag contacts with the inductive load less than 770µH.

Recently, a real-time communication such as digitizing and packetizing voice, video, and fax has become one of the most important services in large computer networks. However, it is hard to support the real-time communication when a failure occurs in large networks. In this paper, management of a real-time large computer network is discussed concerning the link failure. A reliable cluster-based routing algorithm is proposed. The algorithm, which is an application of a cluster based network management, can keep the real-time communication in case of a link failure. By incorporating both the clusters and the multiple routing modes, the algorithm provides not only a short disconnecting time but also small communication overhead of recovery processing. These advantages of the algorithm are shown by the analysis of average message delay based on Kleinrock's model.

We propose a multistage DeBruijn graph, a generalization of the DeBruijn graph, for dense symmetric interconnection networks in multicomputer systems. The proposed graph is symmetric and contains remarkably large number of vertices than other known symmetric graphs such as the Boolean n-cube and the star graph for given degree and diameter.

This paper describes an easy method of generating cloud images by using density contour lines. First, the operator inputs the density contour lines. Features of the contour lines are then extracted. Next, the form is reconstructed on the assumption that it is a homogeneous corpuscle layer of could. Finally, a cloud image is generated, using the reconstructed form and a light attenuation model. The burden on the operator is much lighter than in conventional methods, which use mathematical functions to define objects. Furthermore, the reconstructed shapes can be represented either as mountains or clouds, an important advantage for making a database of computer animation backgrounds. This is an extension of our previously reported method of reconstructing three-dimensional mountainous shapes.

Conventional retrieval methods of image database (IDB) systems are utilizing non-image information, or alphanumerical data, as keywords. These methods are suitable for document database systems, but not for IDB systems because the contents of image cannot be expressed by only such keywords. In this paper, a new image retrieval method using the layout structure" is proposed. This method can realize image retrieval based on the contents of image. Because, the layout structure means structural information such as object location and spatial relation and it corresponds to composition" of the image. This information, called region feature description (RED), is obtained in the input process of original images. In the retrieval process, we use some specified files which are created form RFD on the basis of each structural feature. Thus, interactive image retrieval based on structural information, that is composition", can be performed. Since India ink paintings are adopted as target images in this paper, it is difficult to extract and describe structural feature of object regions as they are. The structural information is described as geometrical feature by approximating them to simple polygons. As a result, composition" of the target image is obtained. This paper describes the description scheme and the extraction procedure of the structural information. And some experimental results are also presented to confirm the effectiveness of this method.