Fault-tolerant synchronisation, data matching and self-starting in redundant systems
Fault-tolerant synchronisation, data matching and self-starting in redundant systems
This dissertation provides economic and practical solutions to the problems of fault-tolerant synchronisation, data matching and self-starting in modular redundant systems. These solutions are based on a simple restriction on the layout of the communication links and easy to implement fault-detection schemes. The level of m-fault-tolerance (the ability to tolerate m simultaneous faults) is achieved using only n ≥ 2m+1 processors and requiring each processor to exchange only a single message per processor per round (a limit that was achieved before only by using n ≥ 3m+1 processors for m-fault-tolerance). Our solutions for the synchronisation problem offer a choice between speed of execution, tightness of synchronisation, and accuracy (closeness to real time). For the data matching problem, the solutions presented offer a choice between serial and parallel data transmission, and between approximate and exact data matching. The solution provided for the self-starting problem allows the non-faulty processors to achieve initial synchronisation with each other irrespective of their initial asynchronism or their initial status (being just switched on, recovering from failure or being introduced to the system to replace faulty ones). A TMR (triple-modular redundant) testbed has been designed and used to test our solutions. The results of testing each of our solutions is presented and discussed.
University of Southampton
1988
Infis, Ali Hadi
(1988)
Fault-tolerant synchronisation, data matching and self-starting in redundant systems.
University of Southampton, Doctoral Thesis.
Record type:
Thesis
(Doctoral)
Abstract
This dissertation provides economic and practical solutions to the problems of fault-tolerant synchronisation, data matching and self-starting in modular redundant systems. These solutions are based on a simple restriction on the layout of the communication links and easy to implement fault-detection schemes. The level of m-fault-tolerance (the ability to tolerate m simultaneous faults) is achieved using only n ≥ 2m+1 processors and requiring each processor to exchange only a single message per processor per round (a limit that was achieved before only by using n ≥ 3m+1 processors for m-fault-tolerance). Our solutions for the synchronisation problem offer a choice between speed of execution, tightness of synchronisation, and accuracy (closeness to real time). For the data matching problem, the solutions presented offer a choice between serial and parallel data transmission, and between approximate and exact data matching. The solution provided for the self-starting problem allows the non-faulty processors to achieve initial synchronisation with each other irrespective of their initial asynchronism or their initial status (being just switched on, recovering from failure or being introduced to the system to replace faulty ones). A TMR (triple-modular redundant) testbed has been designed and used to test our solutions. The results of testing each of our solutions is presented and discussed.
This record has no associated files available for download.
More information
Published date: 1988
Identifiers
Local EPrints ID: 460905
URI: http://eprints.soton.ac.uk/id/eprint/460905
PURE UUID: b127df06-3636-4c90-b6aa-c63d52cd9405
Catalogue record
Date deposited: 04 Jul 2022 18:31
Last modified: 04 Jul 2022 18:31
Export record
Contributors
Author:
Ali Hadi Infis
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics