Markov chains for testing redundant softwareA preliminary design for a validation experiment has been developed that addresses several problems unique to assuring the extremely high quality of multiple-version programs in process-control software. The procedure uses Markov chains to model the error states of the multiple version programs. The programs are observed during simulated process-control testing, and estimates are obtained for the transition probabilities between the states of the Markov chain. The experimental Markov chain model is then expanded into a reliability model that takes into account the inertia of the system being controlled. The reliability of the multiple version software is computed from this reliability model at a given confidence level using confidence intervals obtained for the transition probabilities during the experiment. An example demonstrating the method is provided.
Document ID
19880056148
Acquisition Source
Legacy CDMS
Document Type
Conference Paper
Authors
White, Allan L. (NASA Langley Research Center Hampton, VA, United States)
Sjogren, Jon A. (U.S. Army, Aviation Systems Command, Hampton VA, United States)
Date Acquired
August 13, 2013
Publication Date
January 1, 1988
Subject Category
Statistics And Probability
Meeting Information
Meeting: Annual Reliability and Maintainability Symposium