Failure and Coverage Factors Based Markoff Models: A New Approach for Improving the Dependability Estimation in Complex Fault Tolerant Systems Exposed to SEUs

Author

Ferreyra, Pablo A. ; Viganotti, Gabriel ; Marqués, Carlos A. ; Velazco, Raoul ; Ferreyra, Ricardo T.

Author_Institution

Nat. Univ. of Cordoba, Cordoba

Volume

54

Issue

4

fYear

2007

Firstpage

912

Lastpage

919

Abstract

Dependability estimation of a fault tolerant computer system (FTCS) perturbed by single event upsets (SEUs) requires obtaining first the probability distribution functions for the time to recovery (TTR) and the time to failure (TTF) random variables. The application cross section (sigmaAP) approach does not give directly all the required information. This problem can be solved by means of the construction of suitable Markoff models. In this paper, a new method for constructing such models based on the system´s failure and coverage factors is presented. Analytical dependability estimation is consistent with fault injection experiments performed in a fault tolerant operating system developed for a complex, real time data processing system.

Keywords

failure analysis; fault tolerant computing; probability; random processes; Markoff models; complex fault tolerant computer system; coverage factors; dependability estimation; failure factors; fault injection experiments; probability distribution functions; real time data processing system; single event upsets; time to failure random variables; time to recovery; Application software; Distributed computing; Fault tolerant systems; Operating systems; Performance analysis; Probability distribution; Random variables; Real time systems; Single event transient; Single event upset; Computer dependability models; fault injection; fault tolerance; radiation environment; single event upsets (SEUs);

fLanguage

English

Journal_Title

Nuclear Science, IEEE Transactions on

Publisher

ieee

ISSN

0018-9499

Type

jour

DOI

10.1109/TNS.2007.895548

Filename

4291722