In the future Internet, hierarchically classified Quality of Service (QOS) controls will be effective because various connections requiring different QOS are mixed. However, even in such an environment, among the same class connections, performance protection to harmful impact from the other connections and quality differentiation between connections will be required furthermore. In this paper, from this point of view, we focus on the active connections succession time (age of active connections) as a new dimensional criterion for buffer controls. To be concrete, the packet discarding control of congested router's buffer based on active connections is proposed. Moreover, its performance is evaluated through TCP/IP level simulation from the viewpoint of file transfer time. Conventional Internet can be regarded as the environment where only one class traffic exists (unit class environment). The proposed control scheme can provide powerful differentiation capability to avoid the performance disruption of total connections even in the conventional Internet.

This paper presents a new neural network structure, called Temporal-CombNET (T-CombNET), dedicated to the time series analysis and classification. It has been developed from a large scale Neural Network structure, CombNET-II, which is designed to deal with a very large vocabulary, such as Japanese character recognition. Our specific modifications of the original CombNET-II model allow it to do temporal analysis, and to be used in large set of human movements recognition system. In T-CombNET structure one of most important parameter to be set is the space division criterion. In this paper we analyze some practical approaches and present an Interclass Distance Measurement based criterion. The T-CombNET performance is analyzed applying to in a practical problem, Japanese Kana finger spelling recognition. The obtained results show a superior recognition rate when compared to different neural network structures, such as Multi-Layer Perceptron, Learning Vector Quantization, Elman and Jordan Partially Recurrent Neural Networks, CombNET-II, k-NN, and the proposed T-CombNET structure.

In this article, we first discuss QoS metrics of the data networks, followed by raising the challenging problems for the next-generation Internet with high-performance and high-quality. We then discuss how the WDM technology can be incorporated for resolving those problems. Several research issues for the IP over WDM networks are also identified.

In this paper, we propose a new watermarking method which spreads an ID pattern with specific sequences and embeds it throughout the spatial domain of an image. A set of the sequences is a key for extracting the ID pattern. As an ID pattern is spread throughout an image, we can extract the ID pattern from a part of the image, that is a clipped image. We can also confirm authenticity by extracting the same ID pattern from several parts of an image. The proposed method is robust to StirMark, which is a benchmark tool to test robustness of watermarked images, as well as disturbance by noise addition and common image processing operations such as edge-enhancement, brightness-contrast conversion, posterization, Gauss filtering, median filtering, gamma correction, JPEG compression and clipping.

We consider the problem of embedding complete binary trees into 2-dimensional tori with minimum (edge) congestion. It is known that for a positive integer n, a 2n-1-vertex complete binary tree can be embedded in a (2n/2+1)(2n/2+1)-grid and a 2n/2 2n/2-grid with congestion 1 and 2, respectively. However, it is not known if 2n-1-vertex complete binary tree is embeddable in a 2n/2 2n/2-grid with unit congestion. In this paper, we show that a positive answer can be obtained by adding wrap-around edges to grids, i.e., a 2n-1-vertex complete binary tree can be embedded with unit congestion in a 2n/2 2n/2-torus. The embedding proposed here achieves the minimum congestion and an almost minimum size of a torus (up to the constant term of 1). In particular, the embedding is optimal for the problem of embedding a 2n-1-vertex complete binary tree with an even integer n into a square torus with unit congestion.

Flow control algorithm in high speed networks is a resource-sharing policy implemented in a distributed manner. This paper introduces a novel concept of backlog balancing and demonstrates its application to network flow control and congestion control by presenting a rate-based flow control algorithm for ATM networks. The aim of flow control is to maximize the network utilization for achieving high throughput with tolerable delay for each virtual circuit (VC). In a resource-sharing environment, this objective may also cause network congestion when a cluster of aggressive VC's are contending for the same resource at a particular node. The basic idea of our algorithm is to adjust the service rate of each node along a VC according to backlog discrepancies between neighboring nodes (i.e., to reduce the backlog discrepancy). The handshaking procedure between any two consecutive nodes is carried out by a link-by-link binary feedback protocol. Each node will update its service rate periodically based on a linear projection model of the flow dynamics. The updated service rate per VC at a node indicates its explicit demand of bandwidth, so a service policy implementing dynamic bandwidth allocation is introduced to enforce such demands. Simulation study has validated the concept and its significance in achieving the goal of flow control and yet preventing network congestion at the same time.

Even in today's industries that are predominated by solid-state switching devices, electromechanical devices with electrical contacts are still widely used for switching and/or conveying electrical signals and power. In this paper, some interesting topics in the investigation of electrical contacts, which were selected mainly from those presented at recently held international conferences or submitted to the related Transactions of IEEE and IEICE, are introduced. Specifically, some topics related to investigation regarding contact materials, new techniques for evaluating electrical contact phenomena, new understanding of the contact phenomena, and new applications of electrical contacts are briefly explained.

This paper studies the problem of book-embeddings of graphs. When each edge is allowed to appear in one or more pages by crossing the spine of a book, it is well known that every graph G can be embedded in a 3-page book. Recently, it has been shown that there exists a 3-page book embedding of G in which each edge crosses the spine O(log2 n) times. This paper considers a book with more than three pages. In this case, it is known that a complete graph Kn with n vertices can be embedded in a n/2 -page book without any edge-crossings on the spine. Thus it becomes an interesting problem to devise book-embeddings of G so as to reduce both the number of pages used and the number of edge-crossings over the spine. This paper shows that there exists a d-page book embedding of G in which each edge crosses the spine O(logd n) times. As a direct corollary, for any real number s, there is an ns -page book embedding of G in which each edge crosses the spine a constant number of times. In another paper, Enomoto-Miyauchi-Ota show that for an integer d, if n is sufficiently large compared with d, then for any embedding of Kn into a d-page book, there must exist Ω(n2 logd n) points at which edges cross over the spine. This means our result is the best possible for Kn in this case.

A new method to recover and display 3D fundus pattern on the inner bottom surface of eye-ball from stereo fundus image pair is developed. For the fundus stereo images, a simple stereo technique does not work, because the fundus is observed through eye lens and a contact wide-angle enlarging lens. In this method, utilizing the fact that fundus forms a part of sphere, we identify their optical parameters and correct the skews of the lines-of-sight. Then, we obtain 3D images of the fundus by back-projecting the stereo images.

Scheduling directed a-cyclic task graphs (DAGs) onto multiprocessors is known to be an intractable problem. Although there have been several heuristic algorithms for scheduling DAGs onto multiprocessors, few address the mapping onto a given number of completely connected processors with an objective of minimizing the finish time. We present an efficient algorithm called ClusterMerge to statically schedule directed a-cyclic task graphs onto a homogeneous completely connected MIMD system with a given number of processors. The algorithm clusters tasks in a DAG using a longest path heuristic and then iteratively merges these clusters to give a number of clusters identical to the number of available processors. Each of these clusters is then scheduled on a separate processor. Using simulations, we demonstrate that ClusterMerge schedules task graphs yielding the same or lower execution times than those of other researchers, but using fewer processors. We also discuss pitfalls in the various approaches to defining the longest path in a directed a-cyclic task graph.

A number of parallel algorithms have been developed to solve large-scale real world problems. Although there has been much work on the design of parallel algorithms, there has been little on the design of languages for expressing these algorithms. This paper describes the BPL, a new parallel language designed for butterfly networks. The purpose of this language is to help designers in hiding the complexity of the algorithm and leaving details of mapping between data and processors for lower level. BPL provides a simpler virtual machine for the designer , in order to avoid thinking about control of processors and data. From another point of view, BPL helps designer to logically check the algorithm and correct any possible error in it. The paper gives some examples implemented by this language. In addition, we have also implemented a software tool which simulates the running of the algorithm on the network. The results lead us to believe that this language would be useful in representing all kinds of algorithms on this network including normal algorithms and others.

Abnormal IDDQ (Quiescent VDD supply current) indicates the existence of physical damage in a circuit. Using this phenomenon, a CAD-based fault diagnosis technology has been developed to analyze the manufacturing yield of logic LSI. This method to detect the fatal defect fragments in several abnormalities identified with wafer inspection apparatus includes a way to separate various leakage faults, and to define the diagnosis area encircling the abnormal portions. The proposed technique progressively narrows the faulty area by using logic simulation to extract the logic states of the diagnosis area, and by locating test vectors related to abnormal IDDQ. The fundamental diagnosis way employs the comparative operation of each circuit element to determine whether the same logic state with abnormal IDDQ exists in normal logic state or not.

We propose a learning algorithm for simple deterministic languages from queries and a priori knowledge. To the learner, a special finite subset of the target language, called a representative sample, is provided at the beginning and two types of queries, equivalence queries and membership queries, are available. This learning algorithm constructs nonterminals of a hypothesis grammar based on Ishizaka(1990)'s idea. In Ishizaka(1990)'s algorithm, the learner makes rules as many as possible from positive counterexamples, and diagnoses wrong rules from negative counterexamples. In contrast, our algorithm guesses a simple deterministic grammar and diagnoses them using positive and negative counterexamples based on Angluin(1987)'s algorithm.

Although in recent years, considerable efforts have been exerted on treating the congestion control problems of ABR services in the ATM networks, the focus has been so far mostly on unicast applications. The inclusion of the emerging multicast services in the design of congestion control schemes is still at its infancy. The generic rate-based closed-loop congestion control scheme proposed by the ATM Forum for ABR services suffers from large delay-bandwidth product. VS/VD behavior is therefore proposed by the Forum as an supplement. In this paper, two VS/VD behavior congestion control schemes for multicast ABR services in the ATM networks are examined : forward explicit congestion notification (FECN) and backward explicit congestion notification (BECN). Their performances are analyzed and compared. We further observe that both VS/VD schemes alleviate the problem of consolidation noise and consolidation delay of the RM cells returning from the downstream nodes. The alleviation of consolidation noise and consolidation delay is a major concern of most present researches. Simulation results are also given to support the validity of our analysis and claims.

This paper presents a design and implementation of a ATM multicast service based on programmable and active network concepts. It aims to address the design and implementation issues of creating new network services--multicast in this case--through a set of corba-based network interfaces, and with a java based user codes injection mechanism for supporting customization of network services. We demonstrate the feasibility of our prototype through the implementation of a wavelet video multicast application with active filters implanted at intermediate nodes for supporting heterogeneous receivers and the implementation of a congestion control scheme. The performance of the prototype over an ATM test-bed is measured and evaluated.

This paper describes a unified coding algorithm for lossless and near-lossless color image compression that exploits the correlations between RGB signals. A reversible color transform that removes the correlations between RGB signals while avoiding any finite word length limitation is proposed for the lossless case. The resulting algorithm gives higher performance than the lossless JPEG without the color transform. Next, the lossless algorithm is extended to a unified coding algorithm of lossless and near-lossless compression schemes that can control the level of the reconstruction error on the RGB plane from 0 to p, where p is a certain small non-negative integer. The effectiveness of this algorithm was demonstrated experimentally.

We prove that binary images of irreducible cyclic codes C over GF(2m) and binary concatenated codes of C and a binary [m+1,m,2] even-parity code are optimal (in the sense that they meet the Griesmer bound with equality) and proper, if a root of the check polynomial of C is primitive over GF(2m) or its extensions.

In the domains involving environmental changes, some knowledge and heuristics which were useful for solving problems in the previous environment often become unsuitable for problems in the new environment. This paper describes two approaches to solve such problems in the context of case-based reasoning systems. The first one is maintaining descriptions of applicable scopes of cases through generalization and specialization. The generalization is performed to expand problem descriptions, i. e. descriptions of applicable scopes of cases. On the other hand, the specialization is performed to narrow problem descriptions of cases which failed to be applied to given problems with the aim of dealing with environmental changes. The second approach is forgetting, that is deleting obsolete cases from the case-base. However, the domain-dependent knowledge is necessary for testing obsolescence of cases and that causes the problem of knowledge acquisition. We adopt the strategies used by conventional learning systems and extend them using the least domain-dependent knowledge. These two approaches for adapting the case-base to the environment are evaluated through simulations in the domain of electric power systems.

The use of hand gesture provides an attractive alternative to cumbersome interface devices for human-computer interaction (HCI). In particular, visual interpretation of hand gestures can help achieve easy and natural comprehension for HCI. Many methods for hand gesture recognition using visual analysis have been proposed such as syntactical analysis, neural network (NN), and hidden Markov model (HMM)s. In our research, HMMs are proposed for alphabetical hand gesture recognition. In the preprocessing stage, the proposed approach consists of three different procedures for hand localization, hand tracking and gesture spotting. The hand location procedure detects the candidated regions on the basis of skin color and motion in an image by using a color histogram matching and time-varying edge difference techniques. The hand tracking algorithm finds the centroid of a moving hand region, connect those centroids, and produces a trajectory. The spotting algorithm divides the trajectory into real and meaningless gestures. In constructing a feature database, the proposed approach uses the weighted ρ-φ-ν feature code, and employ a k-means algorithm for the codebook of HMM. In our experiments, 1,300 alphabetical and 1,300 untrained gestures are used for training and testing, respectively. Those experimental results demonstrate that the proposed approach yields a higher and satisfactory recognition rate for the images with different sizes, shapes and skew angles.

Recently, it is required to transfer continuous media over networks without QoS guarantee. In these networks, network congestion will cause transmission delay variance which degrades the quality of continuous media itself. This paper proposes a new protocol using a congestion control with two level rate control in the data transfer level and the coding level. It introduces a TCP-like congestion control mechanism to the rate control of data transfer level, which can detect the QoS change quickly, and adjust the coding rate of continuous media with time interval long enough for its quality. The performance evaluation through software simulation with multiplexing continuous media traffics and TCP traffics shows that the proposed protocol works effectively in the case of network congestion.