An improved reliability model based on software architecture

Applications of existing architecture-based software reliability models are strictly limited to software in which the component transitions satisfy the Markov properties. When software has both deterministic and probabilistic behaviors, these models intentionally model it as Markov process, of which the reliability estimation with lowered accuracy comes as a result. The objective of this paper is to present a new reliability model that can resolve the heterogeneous software architectures which do not fully satisfy the Markov properties. We firstly extend UML sequence diagram to model heterogeneous software architecture. Secondly different types of deterministic component transitions are analyzed. And finally an algorithm generating Markov chain from heterogeneous software architecture is presented. A comparison of reliability estimation result of this new model with existing ones is given with a case study, which indicates a better estimation result could be obtained by this new model.