الفهرس 
AcknowledgmentOnly 14 pages are availabe for public view
 |  | from | 165 |  |  |
| | | from | 165 | | |
Abstract
As the applications of distributed computing systems continue to expand, issues relating to the reliability of such systems are gaining increased attention, and the use of Fault Tolerance in these systems has become inevitable. As a matter of fact, the _mend toward con-strutting such systems has resulted in a two-sided relationship involving reliability and Fault Tolerance. On one hand, it is argued that the multiplicity of processing resources can be exploited to enhance the system reliability. On the other hand, as the number of active elements in a computing system increases, the probability of a failure occuring somewhere at any time also increases, and consequently, efforts must be made to tolerate such faults.
The problem of locating faulty units has been widely studied under a variety of assumptions. In most studies, the system facilities involved in gathering test results and in actually performing the diagnosis were excluded from the system model. Test results are analyzed and a diagnosis is made from what could be called a ”Global Observer” viewpoint. Realizing this unrealistic situation, recent research has been aimed at systems in which diagnosis is performed by the modeled units themselves. In this respect, the diagnosis problem is more complicated