A reliability prediction model for complex systems using data flow dependency

Research on software reliability prediction is of great practical importance. Failure characteristic of large and complex software depends on the operations of individual components and their architecture. Complexity of the components as well as their dependency provides a greater impact on overall reliability of the software. Reliability prediction in the early stage of component based software requires the knowledge of interconnection between the components as well as propagation of errors between the components. In this paper we propose a reliability prediction model which not only considers the control flow of the component it also considers data sharing between the components and their deployment details. We propose a new graphical structure Data Flow Dependency Graph to estimate the effective reliability of the data processed by the components and then use the operational profile to predict the reliability.