Dynamic Data-Driven Fuzzy Modeling of Software Reliability Growth

Abstract

The paper deals with a new model description of software reliability growth dynamics. The model is applicable to multi-stage reliability growth that covers the complex defect detection rate and is based on TakagiSugeno fuzzy inference engine. The identification procedure determines model structure in real time based on evolving clustering algorithm. The clusters are discovered according to Gustafson-Kessel distance metric that copes with clusters of different shape and orientation. The developed model is validated through a case-study data set.