MTTF estimation using importance sampling on Markov models

One of the most basic measures in the logistics process is the MTTF, the Mean Time To Failure. Its evaluation is an important quantitative element in the decision process and it is usually performed using Markov models. This talk focuses oil some difficulties that are found when trying to evaluate the MTTF using Monte Carlo techniques. These are the only available methods that can deal with large and complex models, but they are sensitive to the rarity of the events of interest (for instance; the probability of failure occurrence) Many variance reduction techniques have been proposed to overcome this difficulty. In this talk we make some proposals in the area, and we show their effectiveness and their efficacy.