A non-parametric statistical algorithm is then
developed to estimate the optimal software
rejuvenation schedule, provided that the sample data to
characterize the system failure times is given. It is
important to highlight that the proposed statistical
algorithm is useful in determining the optimal
rejuvenation schedule early in the operational phase; in
contrast, the probability distribution of the system
failure time cannot easily be estimated from a few data
samples. Moreover, the non-parametric approach does
not even require any assumption concerning the
underlying failure time distribution. Finally, the
estimators of the optimal software rejuvenation
schedule have nice convergence properties.