The use of prime implicants in dependability analysis of software controlled systems

The behavior of software controlled systems is usually non-binary and dynamic. It is, thus, convenient to employ multi-valued logic to model these systems. Multi-valued logic functions can be used to represent the functional and temporal relationships between the software and hardware components. The resulting multi-valued logic model can be analyzed deductively, i.e. by tracking causality in reverse from undesirable `top' events to identify faults that may be present in the system. The result of this deductive analysis is a set of prime implicants for a user-defined system top event. The prime implicants represent all the combinations of basic component conditions and software input conditions that may result in the top event; they are the extension to multi-valued logic of the concept of minimal cut sets that is used routinely in the analysis of binary fault trees. This paper discusses why prime implicants are needed in the dependability analysis of software controlled systems, how they are generated, and how they are used to identify faults in a software controlled system.