Architecture-based software reliability modeling

In this paper, we present an architecture-based approach for modeling software reliability. Our approach aims at modeling reliability on various software infrastructures and in any stage of software life cycles. To this end, we utilize characteristics of architectural styles to capture non-uniform behaviors of software embodying heterogeneous architecture. Furthermore, a state model that synthesizes all different architectural styles embedded in the system is developed, allowing the Markov-based reliability model to be employed. Our model can be applied to software with heterogeneous architecture, can facilitate the making of architecture design decision, and is suitable for use in the testing and maintenance phases during which software changes take place. To validate the model, we applied it to an industrial real-time component-based financial system and obtained significant promising results. It is expected that our model have great potential for use to improve software quality effectively.