Showing papers in "Transactions of the Institute of electronics, information and communication engineers in 1993"

Congestion Control and Resource Management in Diverse ATM Environments

Abstract: Large propagation-bandwidthproducts and diverse traffic characteristics have made traditional network congestion control and bandwidth management schemes less effective in high speed networks such as ATM. Furthermore, heterogeneities in ATM network size and the services offered must be considered in the development of effective congestion and resource control disciplines. New mechanisms for traffic and resource control must necessarily be adopted if high speed networks are to operate in an efficient manner. Various congestion and resource control schemes are surveyed and evaluated from the perspectives of high speed local area and wide area networks. They are also evaluated from a service perspective. This material is based upon work supported by the National Science Foundation under Grant No. NCR-8907909. This research is also supported in part by the University of California MICRO program, Nippon Steel Information and Communication Systems Inc. (ENICOM), Hitachi Ltd., Hitachi America, and Tokyo Electric Power Company.

The artificial life of synthetic actors

Abstract: This paper discusses the problem of simulating the artificial life of realistic synthetic actors. It first presents the development of the general concept of autonomous actors reacting to their environment and taking decisions based on perception systems, memory and reasoning. An example of vision-based animation with an individualized walking model is detailed. Then, the authors discuss facial animation techniques for synthetic actors. Finally, they show how to improve the appearance of synthetic actors by dressing them and designing hair styles for them

Methods for Choosing a Training Set which Prevents Over-Learning

Abstract: A drawback of the error-backpropagation method for a multilayer feedforward neural network is overlearning. The authors have already discussed this problem, and obtained necessary and sufficient conditions for over-learning to occur. Using those conditions and the concept of a reproducing kernel, this paper proposes methods for choosing a training set which prevents over-learning. First, the case is discussed in which pure training data are available. Methods are presented for choosing a training set so that a memorization learning does not cause Wiener over-learning. Next, the case in which noisy training data are available is discussed and methods for choosing a training set so that a rote-memorization learning does not cause Wiener overlearning are shown.

A Randomized Algorithm for Finding a Near-Maximum Clique and Its Experimental Evaluations

Abstract: The so-called Maximum Clique Problem is one of the most famous NP-complete problems for which it is difficult to find a solution. Given an indirected graph, we present here a polynomial-time randomized algorithm RaCLIQUE for finding a near-maximum clique. While the basic idea of the algorithm comes from Boltzmann machines, it employs no simulated annealing at all and hence it is simple to control its execution. We have confirmed in experiments for several random and nonrandom graphs with up to 400 nodes that very good solutions can be found efficiently compared with the other conventional algorithms.

Evaluation of Contact Surface Configuration with a Scanning Laser Microscope

Abstract: Arc discharge between contact electrodes brings material transfer and erosion, which sometimes cause serious contact failure. To achieve better reliability and longer life of electromechanical components, characteristics of material transfer and arc erosion of each contact material under different load conditions have to be studied. However, no effective methods for gathering numerical 3D information on contact surfaces have been available. In this paper, the theory and system of a scanning laser microscope is explained, which is confirmed to be useful equipment for this purpose. A confocal optical system and a focus scan memory of the equipment enable us to measure depth of craters and height of pips on contact surfaces in a short time without damaging specimens. By processing the measured data with an image-data processing software, 3D images of contact surfaces are displayed and volume of craters and pips is easily calculated. Relay contact surfaces operated in a capacitive load circuit are numerically evaluated with the scanning laser microscope. The 3D data such as height, depth and volume of craters and pips are measured and material transfer rate per unit arc energy is calculated. The results are shown to confirm how useful the equipment is for investigation of electrical contacts.

An Analysis of Nonreciprocal Phase Characteristics of Magneto-Optic Channel Waveguides Using Finite-Element Method

Abstract: To evaluate nonreciprocal phase characteristics of a magneto-optic channel waveguide, the formulation of a finite-element method based on a scalar wave approximation is discussed here for the first time. This problem is ascribed to the problem of solving a nonlinear eigenvalue equation. However, a simple iterative method is proposed here for solving the nonlinear eigenvalue equation. The validity and usefulness of the present method are demonstrated in analyzing magneto-optic rib waveguides and magneto-optic ridge waveguides. A discussion to increase a nonreciprocal phase shift in those waveguides also is given.

A Method of shape Description by a Distribution Function

Abstract: Some coding methods aimed at data compression for image communication and storage code the image by regions. This paper proposes a method for describing shapes of silhouette figures efficiently with a small amount of data in these coding methods. This method is one which represents the inside regions of a figure by a set of primitive structuring elements, and the primitive structuring elements are used whose contours are defined by a density distribution function. An algorithm, which automatically obtains a collection of primitive structuring elements approximating the contour of an objective region based on the thinned skeleton of the region has been designed. Moreover, based upon the multiple resolution representation of the region obtained by smoothing the contour with the Gaussian filter, the elements in the central part to the elements in the dilated parts are represented hierarchically by a tree structure. Simulations of the descriptive method were done to object regions in a natural image. Since each element has a property of mutually fusing compared to methods which decompose shapes with fixed-shaped elements, this method can approximate the shape of the region well with less data. Moreover, by tree representation of elements, reproduction of the region at various degrees of approximation and resolution becomes possible.

A Multicast Hybrid ARQ Scheme with Stored Error Recovery

Abstract: A multicast error control scheme which takes advantage of both MDS codes and GMD decoding is proposed and analyzed. The proposed scheme is characterized as follows: 1. The sender does not retransmit data frames. The frames to be retransmitted consist of the redundant symbols obtained by encoding the previously transmitted data frames. 2. Each receiver stores erroneous frames not greater than the prearranged number and decodes them all of a lump using the retransmitted frames of the fixed number. Numerical results show that the proposed scheme yields higher throughput not only for the case where the bit error rate is very large but also for the case where the number of VSATs is very large.

SAW Filter Matrix for SS/FDMA System

Abstract: The configuration and function of a surface acoustic wave (SAW) filter matrix used on board a satellite to implement a subchannel switched FDMA (SS/FDMA) system is described. The function and configuration of the SAW filter matrix (SFM) are described. Conditions imposed on the amplitude and phase of two adjacent subchannels to provide a broad channel by combining the outputs of the subchannels are investigated. Electrical and transmission characteristics of an SAW filter matrix used in four-beam SS/FDMA system with a subchannel-combining function are presented, demonstrating its technical feasibility. When implementing an SFM, it is important that the devices used for a satellite-based SS/FDMA router are sufficiently reliable for the design life of the satellite. A methodology is shown that gives a redundant configuration with the lightest mass and that achieves the targeted reliability. Example calculations for a four-beam SS/FDMA router are given.

Recovery of Hierarchical Part Structure of 3D Shape from Range Image

Abstract: A formulation of the part decomposition problem motived by the minimum-description-length (MDL) criteria is presented. Unlike previous MDL approaches which use analytic functions, a general geometric constraint, convexity, is used as a part constraint. Therefore, the method is suitable for complex natural shapes such as human faces. The recovery process consists of a bottom-up grouping process and a subsequent optimization process based on the MDL criteria. The definite causal relations of part structure between different sensitivity levels are used to recover the hierarchy of part structure. Part decomposition experiments involving real 3-D range images are reported.<>

Study on On-Board Large Mesh Antenna Reflector with High Surface Accuracy

Abstract: An increase in the size of on-board antenna reflectors is a key technology in providing economical services by miniaturized portable equipment in future mobile satellite communication systems. In practice, a mesh antenna can be used as the on-board antenna. However, since the reflector surface of this antenna is supported by the cable and the mesh, and its size is larger, the surface is subject to larger thermal deformation in orbit than is the solid antenna reflector. Conventional compensation methods such as passive thermal control or rotation of the antenna reflector cannot maintain the required antenna performances. Various methods have been reported to compensate for the deterioration of antenna performance by thermal deformation. This paper describes a static shape control system for a 10-m diameter antenna which actively compensates for the deformation by using a shape-measuring sensor and shape-compensation actuator. The configuration of the sensor and the experimental results by a functional model are described to confirm the design validity. The shape-compensation method by the actuator was confirmed to be useful in achieving a required surface accuracy.

Accurate Multi-Viewpoint Stereo Measurement Using Nonlinear Optimization Method

Abstract: An estimation method is presented for accurately determining the relationship between viewpoints and object positions simultaneously by integrating stereo data from multi-viewpoints whose relationship is unknown. With this method both the relationship and the object positions are estimated by minimizing the sum of square errors in the image coordinates. The calculation is performed with the nonlinear optimization method because the multi-viewpoint stereo measurement system is a nonlinear system. Therefore, this method can be applied to an arbitrary multi-viewpoint stereo measurement system, whereas almost all conventional methods can only be applied to linear cases. A simulation is used to investigate the influence of the relative viewpoint position parameters on the accuracy with which object positions are determined in a three viewpoint stereo measurement. The effectiveness of this method is confirmed experimentally by actual outdoor measurements.

Topic focusing mechanism for speech recognition based on probabilistic grammar and topic markov model

Abstract: This paper describes a new stochastic topic focusing mechanism for reducing the perplexity of natural spoken languages. In this mechanism, a predictive context-free grammar (CFG} parser analyzes input speech and generates gra mmar-rule sequences. These rule sequences drive a hidden Markov model (HMM), and the current topic is estimated as the HMM state distribution. The CFG rule probabilities are dynamically changed according to this topic state distribution.

Satellite Communication Applications to B-ISDN

Abstract: The terrestrial communication network is making progress from the analog public network based mostly on telephone traffic to the N-ISDN. The development to BISDN also is considered, aiming at an integrated network which can provide large-scale file transfer, high-speed data communication such as the communication between high-speed LAN, and dynamic image communication such as TV and HDTV. This paper proposes the application of the multibeam Satellite communication to such B-ISDN, and presents such technical subjects as the satellite communication. First, the purpose of application of the satellite communication to B-ISDN is described. The requirement in the application to B-ISDN is presented. Then, aiming at the fulfillment of the requirement, the technical subjects and evolution are shown for the configuration of the communication system, the channel control system, the synchronization control system, OAM system, and the interface to the terrestrial network.