Estimating TMR Reliability on FPGAs Using Markov Models

Daniel McMurtrey, Keith Morgan, Brian Pratt, Michael Wirthlin

ScholarsArchive (Brigham Young University) · 2008 · 32 citations · 1 references

Full text

Open access

Concepts

Abstract

A fundamental problem in fault-tolerant computing is pre-dicting a system’s reliability. Several methods exist to model system reliability. Markov modeling, named for the Russian mathematician Andrei Markov, is one such method. The underlying assumption of Markov models is that the probability of a state transition depends only on the cur-rent state. These are called first-order Markov models. Other higher-order Markov models exist, but these are mathemati-cally much more difficult to deal with. The systems analyzed in this paper can all be modeled as a first-order Markov process. Since the conditional probability of being in any state j at time t in a first-order Markov process only depends on the previous state i, the conditional probability of transitioning from one state to the next can be assembled in a square matrix called a transition matrix T. Each entry in the transition

References

1