A neural network approach called the "Gradual Neural Network (GNN)" for the time slot assignment problem in the TDM multicast switching system is presented in this paper. The goal of this NP-complete problem is to find an assignment of packet transmission requests into a minimum number of time slots. A packet can be transmitted from one source to several destinations simultaneously by its replication. A time slot represents a switching configuration of the system with unit time for each packet transmission through an I/O line. The GNN consists of the binary neural network and the gradual expansion scheme. The binary neural network satisfies the constraints imposed on the system by solving the motion equation, whereas the gradual expansion scheme minimizes the number of required time slots by gradually expanding activated neurons. The performance is evaluated through simulations in practical size systems, where the GNN finds far better solutions than the best existing altorithm.

Given the recent advances in telecommunication services, new service operations are needed that support not only existing network services, but also information and communication providing services, such as multimedia services. It is particularly important that service usage information should be gathered, processed, analyzed and presented properly for smooth service provisioning and business support. This paper defines a reference model for service usage information and an implementation model for establishing the functions needed to gather and process this information. The results of an service operation system as implemented in NTT multimedia service trials, currently underway in the city of Urayasu, are also described.

We present a new method for the adaptive control of nonminimum phase continuous-time systems based on the pole-zero placement using approximate inverse systems to avoid the unstable pole-zero cancellations. Using this method effect of the unstable zeros cab be compensated approximately. We show how unstable pole-zoro cancellations can be avoided, and that this method has the advantage of being able to determine an approximate inverse system independently of the plant zeros. The proposed scheme uses only the available input and output data and the stability using approximate inverse systems is analyzed. Finally, the results of computer simulation are presented to illustrate the effectiveness of the proposed method.

This paper discribes a shared medium access control protocol for residential home and small business customer ATM network application with distributed high-speed multimedia services. This protocol offers global fairness for the whole network with a CG (Centralized Grant) signal and, at the same time, offers local fairness between nodes with DG (Distributed Grant) signals. The DG signal is only meaningful to the next neighboring node. The local fairness is kept between two nodes in distributed way, but this distributed local fairness helps to keep the global fairness. Because the DG signal is related to the passed DG signal from the neighbor node. This protocol is rather simple but shows high performance than the already announced protocols.

This paper proposes a prediction method for non-stationary time series data with time varying parameters. A modular structured type neural network is newly introduced for the purpose of grasping the changing property of time varying parameters. This modular structured neural network is constructed by the hierarchical combination of each neural network (NNT: Neural Network for Prediction of Time Series Data) and a neural network (NNW: Neural Network for Prediction of Weights). Next, we propose a reasonable method for determination of the length of the local stationary section by using the additive learning ability of neural networks. Finally, the validity and effectiveness of the proposed method are confirmed through simulation and actual experiments.

This paper presents a discrete-time multiple short-time Fourier transform (MSTFT) suitable for a time-frequency analysis and synthesis of discrete-time nonstationary signals. An overcomplete set of multiple windows in used for a frame constitution in l2 (Z) so that higher quality signal analysis and perfect reconstruction of the signal are achieved. A design method for a prototype window is given where the window can satisfy regularity condition and have a flexible, good time and frequency characteristic under constraint of the uncertainty principle. A dual frame is constructed using the prototype windows in the framework of a frame operator method. Efficient implementation structures for the MSTFT and its inverse transform appropriate for real time numerical processing is presented. Simulation results are given to illustrate the effectiveness for design of the MSTFT. The performance advantages as a new signal analysis tool are demonstrated with an experimental signal.

Neural networks can be used as associative memories which can learn problems of acquiring input-output relations presented by examples. The learning time problem addresses how long it takes for a neural network to learn a given problem by a learning algorithm. As a solvable model to this problem we analyze the learning dynamics of the linear associative memoty with the least-mean-square algorithm. Our result shows that the learning time τ of the linear associative memory diverges in τ (1-ρ)-2 as the memory rate ρ approaches 1. It also shows that the learning time exhibits the exponential dependence on ρ when ρ is small.

This paper describes the analysis of integrated optical waveguides using Finite-Difference Time-Domain (FDTD) method, and proposes the design methodology for low loss waveguide components: corner bends and branches. In order to integrate optical waveguides with Si VLSI technologies on a chip, the compact bends or branches are necessary. Since the optical power radiation from a bend or a branch point depends on the waveguide shapes, an accurate analysis of guided wave behavior is required. For the purpose we adopted the FDTD method which can analyze optical waveguides with a large variation of refractive index and arbitrary shape. Proposed design concept is to have all waveguides transmit only the fundamental mode and to design whole waveguides based on the fundamental mode transfer characteristics. For this design concept, waveguide components are required to have not only low radiation loss but also a little mode shift from the fundamental mode. The bend using the double-reflection mirrors and the branch using a slit are proposed for suppressing the mode shift and improving radiation loss. By the FDTD analysis, the following results have been obtained. The radiation loss and mode shift of double reflection bend are 1% and 4%, and those of the slit branch are 2% and 5%, respectively, in 2 µm width waveguide.

The problem of non-preemptive scheduling of real-time periodic tasks with specified release times on a uniprocessor system is known as NP-hard problem. In this paper we propose a new non-preemptive scheduling algorithm and a new static scheduling strategy which use the repetitiveness and the predictability of periodic tasks in order to improve schedulabilities of real-time periodic tasks with specified release times. The proposed scheduling algorithm schedules periodic tasks by using the heuristic that precalculates if the scheduling of the selected task leads to the case that a task misses a deadline when tasks are scheduled by the non-preemptive EDF algorithm. If so, it defers the scheduling of the selected task to avoid the precalculated deadline-missing. Otherwise, it schedules the selected task in the same way as the non-preemptive EDF algorithm. Our scheduling algorithm can always find a feasible schedule for the set of periodic tasks with specified release times which is schedulable by the non-preemptive EDF algorithm. Our static sheduling strategy transforms the problem of non-preemptive scheduling for periodic tasks with specified release times into one with same release times for all tasks. It suggests dividing the given problem into two subproblems, making a non-preemptive scheduling algorithm to find two feasible subschedules for the two subproblems in the forward or backward scheduling within specific time intervals, and then combining the two feasible subschedules into a complete feasible schedule for the given problem. We present the release times as a function of periods for the efficient problem division. Finally, we show improvements of schedulabilities of our scheduling algorithm and scheduling strategy by simulation results.

This paper optimizes a hierarchical image transmission system based on the hierarchical modulation scheme in a band-limited Rayleigh fading channel. Authors analyze relations between hierarchical parameters and the image quality, and show that the existence of optimum hierarchical parameter that maximizes the received image quality.

A 167-MHz 1-Mbit CMOS synchronous cache SRAM was developed using 0.40-µm process technology. The floor plan was designed so that the address registers are located in the center of the chip, and high-speed circuits were developed such as the quasi latch (QL) sense amplifier and the one-shot control (OSC) output register. To maintain suitable setup and hold time margins, an equivalent margin (EM) design method was developed. 167-MHz operation was measured at a supply voltage of 2.5 V and an ambient temperature of 75. The same margins 1.1 ns of the setup time and hold time were measured for the specifications of a setup time of 2.0 ns and a hold time of 0.5 ns.

A leaky-bucket-with-gate algorithm is proposed to control connection-setup congestion in telecommunication networks providing multimedia services, in place of the call-gapping algorithm used in telephone networks. Multimedia services may use more than one connection simultaneously, while standard telephone services use only one connection at a time. A set of connections used to construct a multimedia service is called a correlated connection group, and the setup requests of such a group form correlated request group. A correlated request group is assumed to be accepted into the network only when all the connection-setup requests for the group are accepted. In this paper, the proposed leaky-bucket-with-gate algorithm, a pure leaky-bucket algorithm, and a call-gapping algorithm are evaluated by simulating traffic with a mix of correlated and uncorrelated connection-setup requests, which models setup requests for video conferencing and telephone services. The simulation results show that the proposed algorithm accepts correlated request groups more efficiently than the pure leaky-bucket and call-gapping algorithms under the simulated traffic conditions, except when the interarrival time in a correlated request group is longer than the acceptance interval. We also present queueing analysis for determining the control parameters in the proposed algorithm. Implementation of this algorithm will facilitate the handling of both setup request traffic for correlated connection groups and for uncorrelated connections in multimedia networks.

The reallocation problem is defined as determining whether products can be moved from their current storehouses to their target storehouses in a number of moves that is less than or equal to a given number. This problem is defined simply and has many practical applications. We previously presented necessary and sufficient conditions whether an instance of the reallocation problem is feasible, as well as a linear-time algorithm that determines whether aall products can be moved, when the volume of the products is restricted to one. However, a linear-time algorithm that generates the order of moving the products has not been reported yet. Such an algorithm is proposed in this paper. We have also previously proved that the reallocation problem is NP-complete in the strong sense when the volume of the products is not restricted and the products have evacuation storehouses show that the reallocation problem is NP-complete in the strong sense even when none of the products has evacuation storehouses.

The key circuit technologies for future giga-bit/low voltage operating high performance SOI-DRAM is described. Emphasis is made especially on the considerations for ways to overcome floating-body effects in order to obtain very long static/dynamic data retention time. A new scheme called a super body synchronous sensing scheme is proposed for low voltage operation at 1 V.

This paper presents severl radio resource scheduling algorithms which aim to provide best-effort service for non-real-time unit-oriented, or message traffic. The objective of resource scheduling algorithm is to distribute radio resources between competing message traffic sources while attaining throughput as high and fair as possible for each source without any explicit quality-of-service (QoS) guarantee. Computer simulations are carried out to evaluate the performance in terms of the average of allocation plus transfer delay, the average of throughput, the variance of throughput, and the usage of resources. The message-size distributions of homepages in World-Wide-Web and e-mails obtained by actual measurement are used. Message size-based resource scheduling algorithms are found to provide high and fair throughput as well as efficient use of the resources.

The performance of synchronous VLSI system is limited by the speed of the global clock which is further constrained by the clock skew. Self-timed design technique, based on the Muller model, improves performance by eliminating the global clock. In order to prevent hazard, a self-timed system should satisfy certain assumptions and timing constraints, therefore special cells are required. The novel Self-timed Cell Library is designed for 1.2µm CMOS technology which contains Muller C-elements, DCVSL circuits, latches and delay elements. It is very useful because: (1) It avoids any possible violations of the assumptions and timing constraints since all cells are custom designed; (2) It provides a fast and reliable model for self-timed system verification using either SPICE simulator or Verilog simulator; (3) It is flexible since it is compatible with an existing Standard Cell Library. In this paper, the library is described. Moreover, the simulated and measured cell characteristics are compared. Using the library, two [18] [81] matrix multipliers employing (1) DCVSL technique, and (2) micropipeline technique have been implemented as design examples and the results are compared. In addition, this paper also demonstrates the benefits of custom-layouted C-elements and a new way to realize delay element for micropipeline. The last but not least, two new HCCs are also proposed.

In this paper we present an algorithm to solve an as-yet untreated knapsack problem, the Multidimensional Multiple-choice Knapsack Problem (MMKP). Since our specific application occurs in the real-time domain, a solution for the MMKP with a small upper bound on the runtime is desirable. Thus, the Lagrange multiplier method is chosen, and a heuristic with a worst-case runtime behavior better than O(n2m) is developed, n being the number of elements and m the number of dimensions. Extensive testing against an exact algorithm based on partial enumeration is used to establish the accuracy and efficiency of the heuristic.

This paper studies stabilization of timed discrete event systems with forcible events. We present an algorithm for computing the region of weak attraction for legal states.

For two decades, our colleagues and we have been developing our multiple learning environments in mathematical education for upper secondary school learners, and have been reporting our learner-centered system in the latest four WCCE Conferences (WCCE/1981/1985/1990/1995). In our latest learning multimedia system, individual learners have to meet a complex network structure in which objectives are arranged in the form of non-linear linking and to proceed actively to their own goals. In order to support their exploring learning, we developed several instructional navigation tools from an instructional view point. This paper presents our instructional navigation technology and its tools. The feature of our present system is to provide a supportive environment where individual learners can set up their own goals, create their own paths for their goals through instructional materials, and construct their own instructional structure based on instructional strategies. This feature is remarkably different from a traditional CAI system in which learners are only directed through the courseware via a linear selection of menus. Also this feature fundamentally differs from general navigation technologies by which a user is able to traverse a series of nodes among non-linear network structure, because our navigation must present individual learners with some easily learnable sequences of objectives based on their object and interest. For this purpose, this system has three chracteristic technologies, i.e. focusing, sequencing and clustering ones. These are very useful for them to make their decisions in order to reach their own goals. This paper consists of (1) ideas of instructional navigation, (2) map technology and (3) navigation technology.

Significant traffic variations occur in telecommunication networks. This paper focuses on hour-to-hour traffic variations during 24 hours and investigates the possibility of using idle periods effectively by introducing non-realtime service at a discount rate. In order to provide non-realtime service, memory is placed at each telecom center. When immediate service is not available, messages of non-realtime calls are stored in the memory and served later when network resources are available. Realtime calls are served with preference. A basic model and a method of dimensioning the memory capacity are presented. The basic model is extended to the network model, and methods of designing a minimum cost network and pricing and profit issues are presented. The results for a basic model are verified by traffic simulation. Numerical examples are given to show the effectiveness of non-realtime service.