Non-parametric Bootstrapping in Software Reliability Assessment
In this paper we consider interval estimations of the software intensity function and its related software reliability measures via the non-parametric bootstrap methods, where the underlying software faultdetection process is described by a non-homogeneous Poisson process with unknown intensity function. We derive the confidence regions of the software intensity function, the expected cumulative number of software faults, instantaneous MTBF, cumulative MTBF, quantitative software reliability as well. The resulting data-driven methodology is based on nine combinations of three bootstrap methods and three definitions of confidence region, and can provide the useful probabilistic information on estimators of several software reliability measures under uncertainty.
Software reliability Non-parametric bootstrap Non-homogeneous Poisson process Kernel-based estimation Confidence region Cross validation
RYOTA HAKOZAKI TADASHI DOHI
Department of Information Engineering, Hiroshima University, Kagamiyama 1-4-1, Higashi-Hiroshima 739-8527, Japan
国际会议
北京
英文
207-213
2011-06-20（万方平台首次上网日期，不代表论文的发表时间）