Reliability Analysis of Dynamic Voting Phased-Mission Systems

Many computing and technological systems can be modeled as a dynamic voting phased mission system (VPMS), which performs a pre-specified task in multiple consecutive phases and the number of components available for use n and the number of components required for the system operation k may vary for different phases. In addition to the changing system configurations, the failure behavior of each component may be phase-dependent due to changing load and operational environments. Furthermore, statistical dependencies across different phases take place for any given component. All those factors contribute to the difficulty and complexity of the VPMS reliability analysis. This paper suggests a new analytical modeling method based on multi-valued decision diagrams (MDD) for efficient VPMS reliability analysis. Unlike existing methods that typically require generating the system reliability model in a phase-by-phase manner and assume identical components and static k-out-of-n structure, the suggested approach encompasses a novel and fast MDD generation algorithm that takes behaviors of all the phases into consideration simultaneously, and can be applied to heterogeneous components with no limitations on their time-to-failure distribution types. Examples of VPMSs with different sizes are provided for demonstrating applications and advantages of the suggested approach.