Diagnosability issues in multiprocessor systems
Abstract
In a seminal paper on fault diagnosis, Preparata, Metze, and Chien introduced a graph-theoretical model. Barsi, Grandoni, and Maestrini relaxed some constraints in this model to create a different model for fault diagnosis. Both these models have become the subject of intense research in the past two decades. A major open problem for these models is the question of sequential t-diagnosability-Given an arbitrary system of units and that there are no more than t faulty units in it, can we always identify at least one faulty unit The author shows that this problem is co-NP complete in both models. Recent research has shown that there are polynomial time algorithms to find the maximum number of faulty units a system can withstand and still identify all of them from a single collection of test results. He presents improved algorithms to solve this problem in both models. Using the letters n,m, and {tau} to denote the number of units, the number of tests, and the maximum number of faulty units respectively, our results can be summarized as follows: in the model of Barsi, Grandoni, and Maestrini, the algorithm has a time complexity of O(n{tau}{sup 2}/log{tau}) improving on the currently known O(n{tau}{sup 2});more »
- Authors:
- Publication Date:
- Research Org.:
- Minnesota Univ., Minneapolis, MN (USA)
- OSTI Identifier:
- 6046358
- Resource Type:
- Miscellaneous
- Resource Relation:
- Other Information: Thesis (Ph. D.)
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; ARRAY PROCESSORS; DIAGNOSTIC TECHNIQUES; FAILURES; ALGORITHMS; FAULT TOLERANT COMPUTERS; MATHEMATICAL MODELS; COMPUTERS; DIGITAL COMPUTERS; MATHEMATICAL LOGIC; 990200* - Mathematics & Computers
Citation Formats
Raghavan, V. Diagnosability issues in multiprocessor systems. United States: N. p., 1989.
Web.
Raghavan, V. Diagnosability issues in multiprocessor systems. United States.
Raghavan, V. 1989.
"Diagnosability issues in multiprocessor systems". United States.
@article{osti_6046358,
title = {Diagnosability issues in multiprocessor systems},
author = {Raghavan, V},
abstractNote = {In a seminal paper on fault diagnosis, Preparata, Metze, and Chien introduced a graph-theoretical model. Barsi, Grandoni, and Maestrini relaxed some constraints in this model to create a different model for fault diagnosis. Both these models have become the subject of intense research in the past two decades. A major open problem for these models is the question of sequential t-diagnosability-Given an arbitrary system of units and that there are no more than t faulty units in it, can we always identify at least one faulty unit The author shows that this problem is co-NP complete in both models. Recent research has shown that there are polynomial time algorithms to find the maximum number of faulty units a system can withstand and still identify all of them from a single collection of test results. He presents improved algorithms to solve this problem in both models. Using the letters n,m, and {tau} to denote the number of units, the number of tests, and the maximum number of faulty units respectively, our results can be summarized as follows: in the model of Barsi, Grandoni, and Maestrini, the algorithm has a time complexity of O(n{tau}{sup 2}/log{tau}) improving on the currently known O(n{tau}{sup 2}); in the model of Preparata, Metze, and Chien, the algorithm has a complexity of O(n{tau}{sup 2.5}) improving on the currently known O(mn{sup 1.5}). He also presents related results in the latter model, which suggest the possibility of reducing the complexity even further. Finally, he develops a general scheme for characterizing diagnosable systems. Using this scheme, he solves the open problem of characterizing t/s and sequentially t-diagnosable systems. The characterizations are then used to rederive some known results.},
doi = {},
url = {https://www.osti.gov/biblio/6046358},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Sun Jan 01 00:00:00 EST 1989},
month = {Sun Jan 01 00:00:00 EST 1989}
}