We propose an asymmetric neural network which can solve inequality-constrained combinatorial optimization problems that are difficult to solve using symmetric neural networks. In this article, a knapsack problem that is one of such the problem is solved using the proposed network. Additionally, we study condition for obtaining a valid solution. In computer simulations, we show that the condition is correct and that the proposed network produces better solutions than the simple greedy algorithm.

It has become more important to reduce the protocol implementation costs as the functions of protocols have become more abundant. The protocol implementation tools which automatically generate a protocol program from a specification described by an FDT (Formal Description Technique) are very promising. Selecting SDL as a target FDT, we have developed an SDL-based protocol implementation tool which consists of a process scheduler and a compiler. Since the efficient SDL process execution is a key to generating the high-speed program, the scheduler is introduced. It provides the mechanism which executes SDL processes concurrently as light-weight-processes. It optimizes so that as few context switches take places as possible. The compiler converts as many kinds of SDL functions whose behaviors can be determined at compile time into programming language statements as possible. These elaborations are so successful that the tool can generate an efficient program. The OSI Transport protocol class 0 program generated by the compiler can process more than 500 packets per second on a 6MIPS workstation.

A novel delivery management system using a new lot sampling scheduling (LSS) method has been developed. The method involves the concepts of "virtual line" and "marker lot," and the system consists of an on-line scheduler executing short-period scheduling for lot-tracking and an off-line scheduler executing long-period scheduling for delivery date simulation. The LSS method can hugely increase the maximum number of lots to simulate the delivery date and also control TAT more effectively compared to conventional dynamic scheduling. Lot progress is controlled by varying the resource allocation ratio for each virtual line. This method is effective for precise delivery date control of lots with various priorities in ASIC production or development lines.

B-ISDN telecommunication systems will require signal processing speeds up to 600 Mbps or more. We must therefore consider the affects of signal reflection, signal attenuation, time dalay, and so on when designing these systems. The higher the signal speed, the larger the electrical noise induced around the connector, especially in the plated through holes (PTHs) area. This paper presents the results of our investigation focused on connector mounting configurations in the signal transmission line, especially whether or not signals transmit through the PTH in a printed circuit board (PCB). How the signal reflection characteristics depend upon transmission line configurations are discussed and experimental results and simulation analyses for a transmission line system using a small miniature A-type (SMA) connector as an example are performed. It is suggested that designs for future high-speed signal transmission circuits take into account the PTH diameter and/or the PTH pitch conditions, values for which can be determined from simulation analysis.

We analyze all-optical switching property of a nonlinear directional coupler (NLDC) having an MQW coupling layer with both nonlinear and linear losses, and examine the effect of nonlinear losses. We use the Galerkin finite element method accompanied by a prodictor-corrector algorithm. The propagation loss along the strongly-coupled NLDC decreases with increasing nonlinear absorption coefficient due to saturation in absorption. A propagation loss of 8.18 dB or 2.38 dB in the bar state of the cross state is much smaller than the bulk loss of MQW structure which exceeds 50 dB. The nonlinear losses lengthen the coupling length and bring it close to that of a lossfree NLDC, while the linear losses shorten. It is found that the property of the cross state is greatly improved by counting the nonlinear losses: The cross-state output power and the output power ratio of two waveguides increase, and the cross state input power, that is, the switching power decreases.

A radial line slot antenna (RLSA) is a high gain and high efficiency planar antenna proposed for DBS subscribers. Spirally arrayed slots are excited by a cylindrical wave with the rotational symmetry. In a small sized antenna where large slot coupling is adopted, aperture efficiency reduction due to rotational asymmetry associated with a spiral arrangement of the slots becomes notable. Authors proposed a RLSA with a concentric slot arrangement excited by a rotating mode in order to enhance the rotational symmetry. This is the first report of the normal operation of a rotating mode RLSA fed by a cavity resonator. The experiments confirm the basic operation of this novel antenna; the gain of 27.8dBi and the efficiency of 68% is measured at 11.85GHz for the RLSA with 0.24mφ.

Broadband antennas such as biconical antennas and log-periodic dipole antennas are generally used in automatic EMC measurements. However, these broadband antennas have not been used for accurate measurement because accurate specifications for them are lacking. Therefore, more accurate analysis is urgently required by the CISPR (International Special Committee on Radio Interference), to establish the specifications for broadband antennas for EMC measurements. In this paper, the AF of biconical antennas is calculated by using Moment Methods. The frequency characteristics and antenna height dependency of AF are presented. AF is also measured and compared to the data obtained by the calculations. Good agreement between the calculations and measurements is achieved, indicating the usefulness of our computation method. In addition, the effect of antenna separation distance and transmitting antenna height on AF is investigated. The calculated AF deviation from the reference value is nearly 0dB except for certain antenna arrangements. In these antenna arrangements, the field becomes null at the receiving antenna and widely varies in magnitude and phase around the null points. Therefore, the antenna is immersed in a non-uniform field, while the AF is defined on the assumption of a uniform field. Consequently, the erroneous AF will be derived from measurements around these null points and it will be greatly different from that obtained at other antenna heights. Thus, it is better to avoid these conditions during actual measurements. The effect of the ground plane on AF is also evaluated. AF is shown to be seriously affected by the ground plane especially at frequencies around 90MHz. It should be noted that AF deviation has crests corresponding to the null field at 300MHz. The obtained data will be useful in establishing specifications of biconical antennas for EMC measurements.

A black membrane is a biological-membrane analogue, i.e. a phospholipid bilayer membrane, artificially formed on an orifice immersed in water. It is used to investigate the properties of the membrane itself and channels embedded therein. In this paper, microfabrication techniques are applied to fabricate the orifice, and a glass substrate is isotropically etched to define the orifice geometry. The periphery of the orifice was patterned with aminosilane to anchor the membrane. The remainder part was coated with fluorosilane to make the surface hydrophobic and to prevent adsorption of channel-forming molecules. We demonstrated experimentally that a stable and reproducible membrane is easily obtainable using the orifice.

For estimating the radiated emission from a metallic enclosure, the authors have developed a numerical computational method which applied inverse analysis. A metallic enclosure containing a loop antenna was set up to be a model source for the numerical analysis. Magnetic fields around the enclosure were measured by measurement systems fabricated in the authors' laboratory. Using the measured magnetic fields, current distributions on the enclosure surface were determined by means of an inverse analysis utilizing the least squares method. From this surface current distribution, the electromagnetic field distributions were estimated by forward analysis on a cylindrical surface 3.0m in radius. The amount of the error in the estimated fields distribution was also discussed.

In this paper, a design of TCM signals for Middleton's class-A impulsive noise environment is investigated. The error event characteristics under the impulsive noise is investigated, and it is shown that the length of the signal sequence is more important than Euclidean distance between the signal sequences. Following this fact, we introduce the shortest error event path length as a measure of the signal design. In order to make this value large, increasing of states of convolutional codes is employed, and the performance improvement achieved by this method is evaluated. Numerical results show the great improvement of the error performance and conclude that the shortest error event path length is a good measure in the design of TCM signals under impulsive noise environment. Moreover, the capacity of class-A impulsive noise channel is evaluated, and the required signal sets expansion rates to obtain the achievable coding gain is discussed.

An efficient algorithm is given for finding all solutions of piecewise-linear resistive circuits containing nonseparable transistor models such as the Gummel-Poon model or the Shichman-Hodges model. The proposed algorithm is simple and can be easily programmed using recursive functions.

Long-wavelength 1.3 µm GaInAsP/InP vertical cavity surface-emitting lasers (VCSELs) have been demonstrated in an array configuration. With the strong current confinement by a buried heterostructure and the efficient optical feedback by a dielectric cavity, five VCSEL elements in a 24 array operated at room temperature with 5 mW total power output and wavelength error within 5%. The stacked planar optics including the VCSEL array is a promising optical transmitter in ultra large scale parallel optical communication systems.

This paper describes the method of applying the integral form of Maxwell's equations to the transmission-line network used in the spatial network method for the modeling of curved conductor surfaces. The techniques of dealing with the transmission-line network near cylindrical conductor surface are explained in detail. To compare exact solutions with computed values, a cylindrical cavity resonator is analysed. The resonant frequencies and unloaded Q's for the computed three modes are obtained with the error of about 1%. Moreover, applying this treatment to the waveguide with magnetron anodeshape cross section, a cutoff-constant is computed successfully. It is found that the treatment proposed in this paper can be used as the method for modeling of curved conductor surface in the spatial network method. It is also considered that this treatment can be extend to TLM method.

It is shown by the derivation of solution methods for an elementary optimization problem that the stochastic relaxation in image analysis, the Potts neural networks for combinatorial optimization and interior point methods for nonlinear programming have common formulation of their dynamics. This unification of these algorithms leads us to possibility for real time solution of these problems with common analog electronic circuits.

In this letter, we demonstrate an experimental CMOS neural circuit towards an understanding of how particular computations can be performed by a T-Model neural network. The architecture and a digital hardware implementation of the learning T-Model network are presented. Our experimental results show that the T-Model allows immense collective network computations and powerful learning.

In the present paper, we attempt to show that the information about input patterns must be as small as possible for improving the generalization performance under the condition that the network can produce targets with appropriate accuracy. The information is defined with respect to the hidden unit activity and we suppose that the hidden unit has a crucial role to store the information content about input patterns. The information is defined by the difference between uncertainty of the hidden unit at the initial stage of the learning and the uncertainty of the hidden unit at the final stage of the learning. After having formulated an update rule for the information minimization, we applied the method to a problem of language acquisition: the inference of the past tense forms of regular and irregular verbs. Experimental results confirmed that by our method, the information was significantly decreased and the generalization performance was greatly improved.

A conventional anode coupled plasma etching process has been developed to etch 300 µm-deep cavities and 600 µm-through holes with nearly vertical sidewalls into single crystal silicon. An optimized SF6/O2 gas mixture results in a nearly vertical etching profile. A silicon wafer was fabricated with a large number of cavities and through holes with less than 1 percent uniformity. It was also experimentally confirmed that this process can be used to etch vertical cavities and through holes in single-crystal silicon with any orientation. This process has the advantage of unlimited etching depth and etching patterns. Advantages in mechanical strength are obtained because a micro-curve is formed at the bottom edge of the cavities. This etching process developed on a conventional plasma etching system was utilized to fabricate a torsional vibrator that consists of single-crystal silicon and Pyrex glass.

Recognition of hand drawn shapes is beneficial in drawing packages and automated sketch entry in hand-held computers. Although it is possible to store and retrieve drawings through the use of electronic ink, further manipulation of these drawings require recognition to be performed. In this paper, we propose a new approach to invariant geometric shape recognition which utilizes a fuzzy function to reduce noise and a neural network for classification. Instead of recognizing segments of a drawing and then performing syntactical analysis to match with a predefined shape, which is weak in terms of generalization and dealing with noise, we examine the shape as a whole. The main concept of the recognition method is derived from the fact that internal angles are very important in the perception of the shape. Our application's aim is to recognize and correctively redraw hand drawn ellipses, circles, rectangles, squares and triangles. The neural network learns the relationships between the internal angles of a shape and its classification, therefore only a few training samples which represent the class of the shape is sufficient. The results are very successful, such that the neural network correctly classified shapes which were not included in the training set.

This paper describes an efficient slew rate enhancement technique especially suitable for an operational amplifier used in an LCD driver IC. This technique employs an input-dependent biasing without directly monitoring an input; instead, monitoring an output of the first stage of the amplifier. This enhancement technique is easily applied to a conventional two-stage operational amplifier and requires only 8 additional transistors to increase slew rates for both rising and falling edges. The bias currents of the first and the second stages are simultaneously controlled by this biasing. Experimental operational amplifiers with and without this enhancement have been fabricated to demonstrate the improvement of slew rate. Slew rates of 12.5V/µsec for the rising edge and 50V/µsec for the falling edge with a 100 pF load capacitance have been achieved by this technique, compared with slew rates of 0.3V/µsec for the rising edge and 5V/µsec for the falling edge in the conventional amplifier.

In this paper, a new off-line handwritten word recognition method based on the explicit modeling of character junctures is presented. A handwritten word is regarded as a sequence of characters and junctures of four types. Hence both characters and junctures are explicitly modeled. A handwriting system employing hidden Markov models as the main statistical framework has been developed based on this scheme. An interconnection network of character and ligature models is constructed to model words of indefinite length. This model can ideally describe any form of hamdwritten words including discretely spaced words, pure cursive words, and unconstrained words of mixed styles. Also presented are efficient encoding and decoding schemes suitable for this model. The system has shown encouraging performance with a standard USPS database.