Assesssing uncertainty in software reliability via quasi-monte carlo methods

ABSTRACT

The need of conducting uncertainty analysis in software reliability for the large and complexity system is demanding. The Monte Carlo method is used for reliability prediction and assessing uncertainty in software reliability. An important improvement of the convergence rate (and thus of speed) can be achieved by using quasi-Monte Carlo methods. These are variants of ordinary Monte Carlo methods, but use quasi-random (highly uniform) sequences instead of pseudorandom sequences. This enhanced uniformity leads to higher rates of convergence. Analysis of a simple problem in software reliability showed that quasi-Monte Carlo methods achieve the same or better parameter estimates as standard Monte Carlo, but have the potential to converge faster and so reduce the computational burden. The paper will explore the use of quasi-Monte Carlo methods to assessing uncertainty in software Reliability.