In the area of system reliability analysis, dynamic and dependent behavior such as functional dependence, cascading failures, and state dependence has been recognized as a significant contribution to problems in overall system reliability. However, with the incorporation of the dynamic and dependent behavior, resulting dynamic system reliability models cannot be efficiently and accurately solved by existing state space based methods such as Markov methods. This is due to two key limitations: the state-space explosion problem, and the inability to handle arbitrary failure distributions. This project explores efficient combinatorial models, formal methods, and transform methods to address these two limitations, providing a viable solution to the accurate reliability analysis of practical large-scale computer-based systems with complex dynamic behavior. The research has four major components: 1) reliability modeling and evaluation of dynamic and dependent behavior, 2) verification and performance analysis of the proposed reliability models and evaluation methods, 3) case studies/industrial applications, and 4) GUI-based reliability software tool development. The new reliability models and evaluation methods developed through this project are fundamental contributions to the body of knowledge on the computer system reliability. This project has broader impact through its contributions to computer system reliability research, graduate and undergraduate education, and the reliable design of complex and dynamic industrial systems. A project website provides publicly available access to the publications generated from the project, software tools and tools-related materials.
