Katsumi Kawano

Autonomous decentralized software structure for integration of information and control systems

A new software structure for integrating information and control systems has been developed. In conventional manufacturing systems, these two kinds of systems are separated and connected by gateways in a hierarchical structure. Both systems need to be integrated, however, in order to increase productivity and to enable the systems to be modified frequently. Timely production and frequent rescheduling of production are required in order to meet rapid changes in customer demands. These integrated systems must also be expandable and flexible. The differences of properties between information and control systems had made it impossible to integrate these systems, so an autonomous decentralized software structure implemented using three new technologies is proposed. These technologies are a timer trigger mechanism that manages the triggering of different kinds of information and control application programs, an interface management mechanism that manages different kinds of data received by these programs, and a message filtering mechanism that segregates the messages transmitted among heterogeneous applications and thus prevents the different applications from interrupting each others operations. This new software structure has been applied to manufacturing systems, and the effectiveness of the proposed technologies has been confirmed.

Phased-in construction method of ATOS

ATOS (Autonomous Decentralized Transport Operation Control System) is a high density railway transport management system of an unprecedented scale. The system has been continuously built up in stages after the online operation of its first station commenced in 1994 and successfully operating since then. In this paper we will discuss phased-in construction techniques by line system. We intend to demonstrate the results of our proving examination of the proposed technology using practical examples. We will also discuss new requirements emerged as the result of improvements made in the network infrastructure for information systems such as information terminals and the extension method thereof. This system, now operating with first and second lines, may be termed the largest of such systems to implement the autonomous decentralization system technology.

The Cooperating Autonomous Decentralized System Architecture

Research can rarely be performed on open platform based large scale distributed systems which always consist of heterogeneous sub-systems and heterogeneous hardware and software components. It is not also pursued to realize the integration of these sub-systems, highly reliable system operations and high productivity of application software development. Our solution is a Cooperating Autonomous Decentralized System Architecture which is an extendible concept of the conventional Autonomous Decentralized System Architecture. The new system architecture is a novel synthesis of four design principles: highly extensible addressing architecture, core communication protocols with minimal semantics, sophisticated network traffic control techniques and reliable communication control mechanism. The current implementation provides both flexibility and high reliability for integrating heterogeneous sub-systems, and it has been proven in several commercial systems.<<ETX>>

An autonomous decentralized system architecture and techniques for on-line development and maintenance

The recent movement to customer-driven production has resulted in the needs for flexible production systems, that are compatible with the localization, the concurrency, and the cooperation of development and maintenance processes under online operation. This paper proposes an architecture and techniques to make online development possible. This architecture enables the online development of frequently reconstructed systems and is based on the autonomous decentralized system concept, which is tolerant of heterogeneous subsystems in a system. Techniques for online development are described, with which development and maintenance subsystems cooperate with online subsystems share mutual data and online data synthetically, and are prevented from disturbing online operation. These features make it possible to develop or maintain systems locally and quickly even when specifications of the whole system are ambiguous. The proposed architecture and techniques have been applied to many industrial production systems, and their effectiveness has been verified in a steel production system.

Autonomous decentralized system test technique

An online system test technique that does not interrupt system operation is proposed. It is based on an autonomous decentralized concept, in which each subsystem has autonomy to control itself and coordinate with the other subsystems. In the autonomous decentralized system architecture, each software module is connected only to the data field (DF), where the data are circulating, and selects whether or not to receive the data on the basis only of the content code of each datum. This means that there exists only one interface between software modules, which is the content code message interface. In this system, both real current data and test data can be circulated in the DF. Hence, a software module can autonomously judge whether to run or test itself on the basis of the received data, while the other software modules are operating. Each module can diagnose the other modules according to their test result data in the DF. The effectiveness of this system test mechanism is shown by applications to real-time control systems.<<ETX>>

Middleware technology for ubiquitous computing: AYA (context-aware and yet another service) that permits autonomous collaboration on super distributed objects

AbstrocrThe present paper proposes a middleware, named AYA, for ad-hoc and autonomous collaboration over the super distributed objects environments where existing information devices cannot be prescribed because of their mobile capability. According to discovery o f the information devices, AYA firstly defines the user context, and secondly assigns roles of information devices based on the scenario for the defined user context, and finally executes the service for the user. These processes are continuously and iteratively conducted where context definition and role assignment in real and dynamic environment are main focuses to be dealt with. Mathematical description of the definition and assignment processes is given as well as some experimental development.

A service selection method based on context types for a ubiquitous service system in a public space

Service systems in public spaces are now being developed by employing ubiquitous computers in these spaces. Such systems must provide services to end users with a wide range of motivations, including unmotivated users who do not actively use the available services. Context-awareness to support these users is thus an important requirement for this kind of ubiquitous computing (UbiComp) system. We propose a service selection method based on registered context types for each end user in order to facilitate ubiquitous service systems in public spaces. By applying this method, we have developed a prototype of an interactive guidance board application and demonstrated its effectiveness.

An autonomous decentralized manufacturing system architecture and its trends for de-facto standards

The recent movement to customer-driven production has generated a need for flexible manufacturing systems that can respond directly to user needs by providing only the volume and type they require, rather than increasing the variety of products. This paper proposes an autonomous decentralized manufacturing system architecture and an information filtering technique for supporting online modification of product lines and free access to production information. It also describes the trends in its industrial de-facto standardization. The proposed architecture has been applied to many industrial production systems, and their effectiveness has been verified.

On-line software test technique based on autonomous decentralized system

A large-scale distributed system is constructed step by step over a long period of time. Even under construction, online operation is started in its system. After starting online operation, it continuously has to be tested and maintained in various areas without stopping online operation. In the conventional online test technique for an autonomous decentralized system, the online and the test data exist concurrently in the same system. The real-time response of the online system is not assured. A method in which test/on-line subsystems are clustered and data flow among each cluster is controlled autonomously, thereby isolating each subsystem and protecting it from perturbation from other subsystems, is given. The reliability and real-time response of the online system are ensured with this method.<<ETX>>

SDO model and its standardization activity

Growth of processor and networking technologies are enabling to connect real-world and computer systems through devices and much attention is being paid to the context-aware services. One of the issues to provide services using ubiquitous or pervasive computing is how to connect heterogeneous hardware device and software component to provide services. This paper proposes a new object model named super distributed objects (SDO) for unified management of them. This model has been proposed to OMG1 and it is adopted by standardization in progress.