Management of control functions in large computer networks is a very difficult problem. One of the effective way to overcome the difficulty is to introduce hierarchical control structure (network cluster) in the management. When a fault occurred in the cluster, routing information at some nodes in the network must be updated in order to react the fault. However, the number of such nodes can be reduced by introducing ingenious topology into the cluster. This paper presents a fundamental discussion on network topology for a network cluster. First, L-FT is defined to represent a degree of fault-tolerance in a cluster with respect to link failures. Secondly, the minimum link problem M is defined to find the minimum number of links to make the cluster L-FT. The following results are obtained. (1) For a network cluster with the fault-tolerant topology 1-FT, at least 2n-2 links have to exist in the cluster where n is the number of border nodes in the cluster. (2) As far as connectivity of the whole network is held, for multiple L link failures in a L-FT cluster, the update of routing information at each node is localized within only the cluster containing the failed links. (3) Several hierarchical networks with fault-tolerant conditions are presented as case studies for a LAN and a MAN.

Electromechanical devices such as relays, switches, connectors and printed wiring boards have shown quick growth along with remarkable progress of electronic products in a recent few decades. For the present, outstanding tasks common to these devices are further downsizing, higher density and broading of bandwidth, on the assumption of keeping high performance and high reliability. These tasks are realized by part production improvement for higher precision and automatic assemblies, development of constituent materials and development of breakthrough structural technologies such as a surface mount technology and an active assembly ferrule technology in optical connectors and so on. In this paper, the technical trends as well as the motive technologies are overviewed for each device.

This paper describes a noise resistant algorithm for estimating 3-D rigid motion from optical flow. We first discuss the problem of constructing the objective function to be minimized. If a Gaussian distribution is assumed for the niose, it is well-known that the least-squares minimization becomes the maximum likelihood estimation. However, the use of this objective function makes the minimization procedure more expensive because the program has to go through all the points in the image at each iteration. We therefore introduce an objective function that provides unbiased estimators. Using this function reduces computational costs. Furthermore, since good approximations can be analytically obtained for the function, using them as an initial guess we can apply an iterative minimization method to the function, which is expected to be stable. The effectiveness of this method is demonstrated by computer simulation.

We predict that chemical waves can propagate as a guided mode in a reaction-diffusion system that consists of two regions with different wave speeds. In comparison with electromagnetic waveguides, unique features of the guided chemical waves can be seen in their dispersion characteristics. Conditions for supporting lowest-loss guided waves are discussed.

Nonmonotonic reasoning is a logical inference system which attempts to approximate human commonsense reasoning and is characterized as defeasible: having reasonably drawn a conclusion from some premises we may be forced to retract that conclusion upon learning new facts. This paper introduces a Petri net model for nonmonotonic reasoning with nonmonotonic rules generated by annotated logic programs and the unless operator. In the Petri net model, a fixpoint of a nonmonotonic theory can be represented as a maximal and consistent support of a firing sequence. We propose a structural method for finding extensions (coherent consequences) for a given set of nonmonotonic logic rules. It is based on the T-invariant technique for testing fireability of a goal transition in the Petri net model of Horn clause logic programs.

State space explosion is a serious problem in analyzing discrete event systems that allow concurrent occurring of events. A new method is proposed for generating reduced state spaces of systems. This method is an improvement of Valmari's stubborn set method. The generated state space preserves liveness, livelocks, and terminal states of the ordinary state space. Petri nets are used as a model of systems, and a method is shown for generating a reduced state space from a given Petri net.

As a result of examination based on a newly available data set of millimeter-wave rain attenuation measured in the UK, it is found that the ITU-R specific rain attenuation model tends to appreciably underestimate millimeter-wave rain attenuation at frequencies above about 60GHz for the UK rain climate. This tendency is very similar to that previously reported for the Japanese experimental data at frequencies up to 245GHz. Furthermore, an alternative specific rain attenuation model based on the Japanese experimental data is found to be in fairly good agreement with the experimental data in the UK at frequencies up to 137GHz.

The prediction of a relay contact's life is still very important for support and maintenance of the Crossbar Switching Systems. It was found through surveys and experiments that the protected shower arc is the main reason for switching-relay contact erosion at existing Crossbar Switching Systems, if the contacts were not heavily activated. If the contacts were heavily activated, a long sustained steady arc might occur and severely erode the contacts. This paper proposes an arch energy estimation method for the prediction of contact erosion using object-oriented simulation technology when a steady arc occurs at protected contacts. The arc energy is expressed in a simulation model through analysis, and the model was confirmed through experimentation. The simulation model was used for building block programs of an expert system to predict the life span of switching relays in the existing Crossbar Switching Systems.

This paper describes the optical characteristics and static fatigue reliability of a zirconia alignment sleeve, which is a component part of an optical connector with zirconia ferrules. This combination of sleeve and ferrules hardly generates any wear debris during connector insertion and removal cycles. This has reduced the cleaning frequency of the ferrule endface during cycles and greatly improved the return loss stability of the optical connectors. The zirconia alignment sleeve enables stable return loss characteristics to be achieved over a wide temperature range as it has the same thermal expansion coefficient as the zirconia ferrule. Furthermore, the gauge retention force for the zirconia alignment sleeve is defined with a view to its practical use. This force must be between 2.0 and 3.9 N to allow stable optical connections to be made under various mechanical and environmental conditions. We also clarify the conditions for a proof test by which to prevent the occurrence of static fatigue fractures in the sleeve, and we confirm the validity of the test. The static fatigue parameters for zirconia ceramics and derived from the static fatigue theory for brittle materials and fracture testing. We use these static fatigue parameters to predict the lifetime of a zirconia sleeve under working stress. An appropriate stress level for the proof test which eliminates weak sleeves, is about 3 times greater than working stress. The strength of the sleeve as demonstrated in the proof test is confirmed by accelerative stress aging. The performance of this sleeve is superior to that of a conventional copper alloy sleeve and the proof test confirms its reliability; under 0.1 FIT for 20 years of use.

We analyze the effect of overlapping technique on differential pulse position modulation (DPPM) in optical direct-detection channel when the pulsewidth and the average power of the channel are constrained. We refer to the modulation scheme employing an overlapping technique in DPPM as differential overlapping PPM (DOPPM). To avoid frame synchronization problems, we analyze the performance of DOPPM under the window scheme that results in lower bounds on the capacity and the cutoff rate of DOPPM but is easy to analyze. Under this scheme, we analyze the lower bounds on the capacity and the cutoff rate of DOPPM. It is shown that DOPPM with the window scheme has higher capacity and cutoff rate than PPM and DPPM, and also than OPPM when the average received number of photons per slot is somewhat large. The overlapping technique is thus shown to be effective on DPPM under the pulsewidth constraint when the average received number of photons per slot is somewhat large.

High-speed pulse propagation, up to several hundred Mbps or higher, will play an important role in telecommunication systems for B-ISDN. High-performance packaging, especially high-speed, high-throughput interconnection, is strongly required. For advanced telecommunication systems, giga-bit signal transmission has been developed at the multi-chip module level, and 300 to 600 Mbps signal transmission has been reached at the printed circuit board level. Electrical inter-cabinet interconnections of 150 to 300 Mbps have been achieved for up to several tens of meters. High-speed, high-throughput connectors are the key to achieving high-performance telecommunication packaging systems. Two technologies are extremely important. One is for high-density, high-pin-count connectors, and the other is for high-speed signal transmission connectors. The requirements for the connectors needed for advanced high-performance telecommunication systems are described. Several high-density, high-bandwidth connectors developed for high-performance packaging system are introduced.

The sidelobe canceler in radar systems is a highly computational demanding problem. It can be efficiently tackled by resorting to the QR decomposition mapped onto a systolic array processor. The paper reports several mapping strategies by using massive parallel computers available on the market. MIMD as well as SIMD machines have been used, specifically MEIKO Computing Surface, nCUBE2, Connection Machine CM-200, and MasPar MP-1. The achieved data throughput values have been measured for a number of operational situations of practical interest.

This paper describes a trial coaxial surface mounted connector for PGA-type high-speed multichip modules (MCM). An MCM connector is needed to ensure testability and connection reliability of MCMs mounted on a printed circuit board. Our connector consists of a coaxial elements, a common ground housing made of conductive resin, and a ground contact spring plate. It has 68 signal contacts. We investigated the performance of this connector by experiment and simulation. Its insertion force is only about 53 gf per signal pin. The characteristic impedance is from 45.6 Ω to 61.4 Ω. The average resistance between two contacts is 28 mΩ with a deviation of less than plus or minus 5 mΩ. The insertion is -0.4 dB at 1.0 GHz. Crosstalk noise is less than 1.2%. This prototype connector can transmit pulses of up to 1.2 Gb/s, showing that it is applicable to high-speed MCMs.

The electric or electronic circuits have many contact devices such as relay and switch. The contact between two nominally conducting flat surface has a lot of micro contact spots. The constriction resistance of the contact is known to determine the sum of the parallel resistance of the micro contacts and the interaction of them. The constriction resistance of two circular conducting spots was approximately formulated by Greenwood. This formulation shows that the interacted resistance of two circular spots is in inverse proportion to the distance between two conducting spots. It was known that this effect is introduced by the interaction between two conducting spots. However, the condition of interaction in the spots is not clear. Calculating the current density distribution in the spots is important to clarify the condition of interaction. The numerical analysis is very suitable to calculate the current density in the spots. In the fundamental case of the computation of the current density the boundary element method (BEM) is more efficient and accurate than that of the finite element method (FEM) because the boundary condition at the infinite is naturally satisfied and is not required a great number of the element in a wide space. In this paper the current density in the square spots is computed by the BEM. As the distance between two conducting spots becomes small, the current density in the two spots decreases. It becomes clear that the constriction resistance of conducting spots is increased by this effect. The decrease of current density by interaction is not uniformly, that at the near location to the opposite spot is larger than that at the far location in the same spot. In this paper the constriction resistance of two conducting spots is also considered. It was known that the constriction resistance of one conducting spot is not influenced by the form of spot very much. However, that of two conducting spots is not clear. The constriction resistance of two square spots is also computed by the BEM. The computed values of the constriction resistance of two square spots are compared with that of two circular spots by Greenwood's formulation and other results. As the result, it is clear that they have the considerable discrepancy. However, the trend of the variations is almost agree each other.

Composite materials of solid lubricants, such as graphite, MoS2, WS2, etc., and metals are being used as the sliding electrical contacts. However, few reports have so far been presented on the detailed characteristics of such composite materials. It is shown in this report that contact resistance and coefficient of friction of the sliding contact of the composite material of Cu-Nb system against Cu were higher than those of the sliding contact of the composite material of Cu-Sn system against Cu. It was, further, found that composite materials of Cu-Sn system were superior to those of Cu-Nb system being both contact resistances and coefficients of friction lowered. At the same time, it was found that performances of composite materials of Cu-Sn alloy base containing exclusively WS2 were superior to those containing both WS2 and MoS2. It was, therefore, suggested that proper samples suitable for the service conditions should be selected from the composite materials of Cu-Sn system which contain exclusively WS2 for the practical applications.

We modified the adaptive fuzzy classification algorithm (AFC), which allows fuzzy clusters to grow to meet the demands of a given task during training. Every fuzzy cluster is defined by a reference vector and a fuzzy cluster radius, and it is represented as a shape of hypersphere in pattern space. Any pattern class is identified by overlapping plural hyperspherical fuzzy clusters so that it is possible to approximate complex decision boundaries among pattern classes. The modified AFC was applied to recognize handwritten digits, and performances were shown compared with other neural networks.

Several distances between trees have been proposed. However, most of the reports on distances have dealt with rooted and ordered trees. This paper proposes two distances between unrooted and cyclically ordered trees (CO-trees) and their computing methods. A CO-tree is a tree embedded in a plane. These distances are defined based on Tai's mapping (TM) and a strongly structure preserving mapping (SSPM) between CO-trees. The time complexities to compute the distances between two CO-trees Ta and Tb are OT (N 2aN 2b) for the distance based on a TM and OT(mambNaNb) for that on an SSPM, respectively, where ma(mb) and Na(Nb) are the largest degree of a vertex and the number of vertices of Ta(Tb), respectively. The space complexities of both methods are Os(NaNb). Those distances can be applied to the clustering of CO-trees.

As a generalization of the tree automaton, tree automata with various types of memory are introduced and their relation to context-free grammars with memory is studied. Relations between computation trees of tree automata with memory and derivation trees of context-free grammars with memory are established, and as a consequence, the languages generated by context-free grammars with memory are characterized in terms of the sets of trees recognizable by tree automata with memory. Also various types of traversal of labeled trees recognizable by tree automata with memory are considered.

ISO and ITU-T have developed the standard concept of the Telecommunications Management Network (TMN). This standard does not, however, specify interface implementation. The user requires an Application Programming Interface (API) that bridges user application and the TMN concept to construct a TMN-based application. This paper proposes an object oriented API (OOAPI) that is suitable for TMN-based operations system implementation. OOAPI is one interface of the Common Management Information Service Element (CMISE), and uses the Common Management Information Protocol (CMIP). OOAPI is composed of two C++ programming language constructs: Data Object and Interface object "M_User". The Data Object makes it easier for the user to access management information. The M_User provides a connection-less CMISE interface because the OOAPI handles CMIP association automatically. The M_User also provides MO location-transparency by using the OSI Directory Service. This paper compares the existing MO location-transparency schemes with the OOAPI method, and clarifies the advantages of OOAPI. This paper also indicates results from OOAPI trials, and confirms that OOAPI has sufficient performance to implement highly effective TMN operations system.

Arc discharge between electrodes of relays and switches often causes contact surface damage through material transfer and arc erosion. Especially, material transfer sometimes occurs and brings serious failure even under lower load that is quite smaller than the minimum arc current value of contact material. In this paper, contact surface configuration, material transfer, and arc erosion characteristics of Ag and AgPd 60 contacts were experimentally studied after 0.5 or 1 million switching operations at various load levels. The followings can be made clear. Firstly, it was confirmed that the arcs and material transfer occurred even under such current that was lower than the minimum arc current. Therefore, the definition of the arc occurrence boundary current was newly determined. Secondly, the relation between load conditions (current and power supply voltage) and contact surface configuration (craters and pips) caused by material transfer was studied. The arc erosion behaviors of tested samples could be classified into two types: material transfer type and wear-out type. As one of the primary factors of transition from the former type to the latter one, contact activation was considered. The influences of load conditions and organic gas emitted from relay structure on arc characteristics was experimentally examined. The results indicated that load current greatly influenced the amount of material transfer and that power supply voltage affected the occurrence of the wear-out type significantly. The activation behavior of the contact surface could be found through observing the bridge voltage waveform.