A reconfigurable array for solving linear system of equations (LSE) by LU-decomposition method is considered. It has nearly a half of the number of processing elements (PE) and I/O terminals with reduced computation time compared with known conventional array structures.

A Bayesian technique was used to obtain point estimators for the parameters of a bivariate exponential distribution associated to a two-component parallel electronic system and a point estimator for the system reliability function.

Transverse mode characteristics of a high power 1.3µm GaInAsP/InP buried crescent laser by the use of a dry etching process is described. With this technique, a single transverse mode (STM) yield as high as 95% was achieved. Transverse mode stability was confirmed after 1000 h aging at 50, 20 mW. A coupling efficiency into a single mode fiber as high as 63% was obtained.

In order to calibrate fiber optic reflectoreters accurately, the imperfection of a high-reflection standard of which power reflection coefficient is close to unity has been estimated. The power reflection coefficient of 0.98 has been attained.

In this paper, chaotic phenomena in a simple circuit which has a new type of vector field, are discussed. It has only one nonlinear diode which operates as a limiter, and a real eigenvalue when the diode is on, is equal to zero. In addition, the physical mechanisms of the generation for chaos are clearer. The switching operation of the diode changes the oscillation frequency in the circuit considerably and this ettect causes a stretching and folding mechanism. Moreover by idealizing the diode, the attractor is proved to be chaos in the sense of Li-Yorke. Lastly the theoretical results are verified by circuit experiments.

In this paper, we propose to adopt the asymmetric error correcting code for photon communication systems. The asymmetric error correcting code is an binary code correcting only 10 type transition errors (asymmetric errors). We show the following advantages obtained by employing the asymmetric error correcting code:(i) the codeword error probability is smaller than that of the symmetric error correcting code. (ii) the information rate per photon is larger than that of the symmetric error correcting code. Moreover, for asymmetric error correcting codes, we obtain the lower bounds on the codeword error probability and the upper bounds on the information rate per photon. By using these bounds, we can show that some asymmetric error correcting codes are optimum for these criteria.

An evaluation method is developed which is applicable to teachers' training from the view point of speaking skill as teacher's behavior. Speaking skill mentioned here can be refered to phonetic expression technique or a way of speaking" as a more ordinal expression. In general, as reading a passage with or without intonations makes difference in listeners' impression, we select the evaluation terms regarding the point. The development is done as follows. (1) The speeches of experienced teachers and several trainees were recorded in the digital method. (2) Letting subjects listen to the records, selecting terms on each of which the speech of experienced teacher differs from that of trainees, then we regard each term as evaluation item. (3) In teachers' training, it is necessary to feedback aggregated result of the evaluation immediately and let the trainee know it. So, we adopt radar chart from the viewpoint of easiness of looking at and express the result on it visually. Then we discuss an example of the administration of the evaluation above using the method.

In actual dimensioning of telephone networks, the number of circuits in a link takes only a nonnegative multiple of a certain positive integer M, called the module size". Network dimensioning with this modularity constraint is called modular engineering or modular dimensioning. This paper analyzes phenomena peculiar to modular dimensioning, and proposes a modular dimensioning algorithm. The proposed dimensioning algorithm is based on an efficient search of the solution space, i.e., implicit enumeration approach. Effective rules which take account of the characteristic of this network design problem are used to reduce the amount of searching in the implicit enumeration approach. The proposed algorithm is evaluated on a multiple-triangle network, is compared with existing simple modular engineering algorithms, and is shown to reduce the total link cost by an average of 1-5% lower than the existing algorithms.

A switching network construction scheme for a fast packet switching system, which is expected to provide multi-media services efficiently and economically, is proposed. The proposed packet switch has RAM as a shared buffer. Its access is assigned to input and output ports dynamically in order to fully utilize limited memory time. This enables packet switches to flexibly accommodate packet links having different bit-rates. An asynchronous data transfer technique is also incorporated into inter-switch links. This switching network can dynamically alter the inter-switch link speed, according to the demanded flow. The dynamism of this simple hardware contributes greatly to better packet transfer quality. This paper describes switch architecture, multi-stage packet switching network construction which can handle any combination of packet flow without centralized switch resource management, and simulated packet transport characteristics of a multi-stage switching network. The simulation results show that the proposed switching network gives good packet transport quality, in terms of packet delay and loss. Finally, the congestion mechanism is analyzed.

A large scale voice store forward system has been developed for NTT's ISDN, the Information Network System (INS). This system, which is based on digital signal processing (DSP), offers message delivery, message board and information proffer services, and it has user-friendly advanced features. Single chip DSP LSI are utilized for voice code conversion with pause-in-speech detection and DTMF reception, and a new backward-hangover processing technique has been introduced for pause-in-speech deletion. The field trial showed that the system performed as designed, that the user-friendly features truely helped the user to use it easily, and that the amount of storage used was reduced to 35% by adopting adaptive differential PCM coding and pause-in-speech deletion without voice quality degradation. The system was improved to meet network requirements, and commercial service was started in November 1986.

Electromagnetic wave generation due to electrostatic discharge from charged human body to an earthing electrode has been observed by the use of a pocketable size electrostatic discharge detector. A subject standing on an insulating platform to avoid additional stray capacity has been charged up to 13 kV of positive or negative voltage by a dc power supply. Electrostatic discharge from negatively charged human body has resulted more intensive electromagnetic wave than that from positively charged one. Magnitude of electromagnetic wave has depended on an inserted discharging resistance between the esrthing electrode and the ground. Electromagnetic wave detection rate has been defined, then V₅₀; 50% electromagnetic wave detection voltage has been derived. The experimental results claims that a certain fraction of the power dissipated at electrostatic discharge should be radiated as Poynting vector of radiative electromagnetic noise.

The single-row routing approach to the design of multilayer printed wiring boards decomposes the general problem into a number of independent phases. The layering phase is considered in this paper. We assume that the street capacity is exactly two in each layer. Recent advances in the technology of microelectronics, which enables us to produce high density boards, show that it is possible to drill one via between any two adjacent vertices on the single-row where at most two wire segments are passed. We call these vias as additional vias. We give a new upper bound to the number of layers while using additional vias is permitted. A polynomial time algorithm SPLIT is given for assigning nets to m layers where m is less than or equal to the new upper bound. The number of additional vias is not limited in algorithm SPLIT. For having a reasonable number of additional vias on the single-row, a heuristic algorithm LIMIT_SPLIT is proposed. LIMIT_SPLIT is implemented and compared with a related conventional algorithm. Experimental results show that LIMIT_SPLIT takes the shorter run time and provides smaller number of layers for 59% of the randomly generated net lists while using no more than two additional vias in each case.

This paper gives another consideration for Benes rearrangeable IN to be constructed from cascading two blocking Interconnection Networks (IN), in an attempt to make such type of IN more practical for parallel systems. The reverse binary n-Cube, and binary n-Cube multi-stages INs, have been used to construct rearrangeable Benes IN. The control algorithm has two construction phases. The first phase is to confirm and transform any arbitrary permutation into the conflict free form (to control the reverse binary n-Cube IN). The second phase uses the mechanism of Bit-Tag control to map such conflict free permutations and control the second half of Benes IN, (i.e., the Cube). The second phase receives its input from the first phase output, after being transformed into a conflict free permutation. Such construction increases the flexibility in the design of Rearrangeable Multi-stage IN.