Software Reliability Growth Model`s Assumptions in Context of the Secondary Faults
Dmitry A. Maevsky, Elena J. Maevskaya, Lyudmila Shapa · 2017
The article describes the results of analysis of four general assump- tions of all existing software reliability growth models: the assumptions about similarity of testing and operation conditions, immediate detected fault elimina- tion, the assumption that secondary faults are not brought in and the one of mu- tual independence of faults. Under the real conditions of Software testing and operation none of these assumptions is executed. Because of this fact, all exist- ing models possess low modeling accuracy. The authors have shown that the removal of only one of these assumptions – the one that secondary faults are not brought in – allows to automatically remove all other assumptions. A new Software reliability model created with considering the possibility to bring in secondary faults has demonstrated stably better results as compared to the tradi- tional ones. This model is based not on the theory of probability, but on the the- ory of nonequilibrium processes.