Fault-tolerant real-time guarantees in high speed networks /
This dissertation examines issues concerning fault-tolerant real-time communication in high speed networks such as those used in embedded mission-critical systems. Embedded mission-critical systems like the ones used in military applications are often built around high-speed backbone networks like...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1995.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=742744331&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | This dissertation examines issues concerning fault-tolerant real-time communication in high speed networks such as those used in embedded mission-critical systems. Embedded mission-critical systems like the ones used in military applications are often built around high-speed backbone networks like FDDI and ATM. These systems impose stringent performance requirements on the underlying network. For example, Navy's SAFENET specification requires that the network must guarantee delivery of critical messages by their deadlines in normal as well as some faulty conditions. Designing and managing such networks is a challenging task. The range of issues to be addressed include architecture design for survivability, resource allocation for guaranteeing timely delivery of messages, and development of effective testing methods that can provide such a-priori guarantees. This work will present proposed solutions to these important problems. In the bulk of the thesis, we will illustrate our solutions and methodology in the context of FDDI-based networks. Later, we examine how the approach can be extended to other networks. We, first concentrate on providing real-time message delivery guarantees in an FDDI network in the absence of faults. Next, we consider how the FDDI architecture can be extended to provide a high level of fault tolerance. Finally, we solve the integrated problem of guarateeing timely delivery of critical messages in spite of faults. In comparison to previous work, this thesis makes two significant contributions. First, we propose and analyze a new metric called Guarantee Probability to evaluate the average case performance of hard real-time systems. While worst case metrics for such systems have been studied in some depth, this is the first study on average case performance. Second, we combine our fault-tolerant architecture design with the real-time communication guarantees to provide an integrated solution. |
|---|---|
| Item Description: | Vita. "Major Subject: Computer Science". |
| Physical Description: | xiii, 195 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilms Inc. |
| Bibliography: | Includes bibliographical references. |