Yasukuni Okataku

Development of Intellectual Distributed Processing System

Abstract The Intellectual Distributed Processing System (IDPS) does not have any centralized managing element. IDPS system elements find their roles autonomously, in order to complete given objectives. IDPS system elements are also able to solve conflicts among different objectives without any supervisory element, by exchanging information among elements. In this paper, fundamental mechanisms for managing system elements distributedly are described. These mechanisms are implemented on the object oriented distributed operating system, which has been developed on multi micro computers connected by local area network. It is expected that large scale adaptive systems can be developed and operated easier by these mechanisms.

Power system voltage control by distributed expert systems

Voltage control by distributed expert systems has been investigated as a case study of the application of a distributed control scheme to power system control. The combined injection of VAr-compensating devices controlled by distributed expert systems has been proposed as a measure to maintain voltage stability in heavy loading conditions of a power system. A simulation study has been carried out using five workstations which represent a power system and four VAr-compensating devices. The results demonstrate the effectiveness of the proposed system for voltage recovery.<<ETX>>

A fault-tolerant architecture based on autonomous replicated objects

The paper proposes an architecture for the replication of program modules enabling them to behave in accordance with their own local knowledge, without any influence by not only their location, replication degree and fault-tolerant mechanism but also system level modules. In the proposed architecture, program modules are implemented as objects and communication among them is carried out by a total ordering broadcast protocol, enabling individual objects to behave autonomously. Therefore, individual objects can choose the most suitable replication degree and fault-tolerant mechanism in accordance with their own required reliability and execution efficiency without the need to change programs or for the object location to be influenced.<<ETX>>

A Fault-Tolerant Architecture for the Intellectual Distributed Processing System

The Intellectual Distributed Processing System (IDPS) is a system architecture to realize highly extensible and adaptive distributed systems. The reliability of the IDPS relies on replicated objects model, and this paper presents two new mechanisms for implementing that model on the IDPS. One is a fail-stop broadcast communication protocol, where replicated objects receive the same messages in the same order. The other is a commitment method, where each object receives only correct messages. By these mechanisms, an individual object does not need to be aware of the replication degree and the location of relevant objects. Moreover, each object can communicate with other individual objects without confirming message transmission. Therefore, the overhead for the fault-tolerant mechanisms can be maintained at a small level.

Intellectual Distributed Processing System Development

The Intellectual Distributed Processing System (IDPS) is a system which has no centralized managing element. IDPS system elements find their roles autonomously in order to complete objectives given to the system. IDPS system elements are also able to solve conflicts among different objectives without any supervisory element, by exchanging information among elements.

Distributed Deadlock Avoidance and Detection in Intellectual Distributed Processing System

Distributed systems have advantages in developing large, flexible, and reliable systems. The Intellectual Distributed Processing System(IDPS) eliminates all central managing mechanisms from systems in order to make the best use of these advantages. This paper describes the deadlock avoidance and detection method adopted in the IDPS. The proposed mechanism has the following characteristics. 1) The proposed mechanism, which is completely distributed, is carried out by communication among jobs and resources. Therefore, it is possible to make highly flexible and reliable systems easily, Moreover, the proposed mechanism does not trace individual hold-claim and/or hold-wait relations in directed graphs sequentially. Therefore, it is possible to execute deadlock handlings in parallel on different sites. 2) It is possible to use deadlock avoidance and detection mechanisms simultaneously. Deadlock detection is used for data resources and deadlock avoidance is used for physical resources.