Efficient Analysis of Systems with Multiple States

A multistate system is a system in which both the system and its components may exhibit multiple performance levels (or states) varying from perfect operation to complete failure. Examples abound in real applications such as communication networks and computer systems. Analyzing the probability of the system being in each state is essential to the design and tuning of dependable multistate systems. The difficulty in analysis arises from the non-binary state property of the system and its components as well as dependence among those multiple states. This paper proposes a new model called multistate multivalued decision diagrams (MMDD) for the analysis of multistate systems with multistate components. The computational complexity of the MMDD-based approach is low due to the nature of the decision diagrams. An example is analyzed to illustrate the application and advantages of the approach.