System Safety Requirements as Control Structures

Abstract

Along with the popularity of software-intensive systems, the interactions between system components and between humans and software applications are becoming more and more complex. This results in system accidents related to system safety issues. System accidents are different to failures related to component reliability. System safety is not well addressed, because functional requirements and safety requirements are separately handled in practice. In this paper, we consider safety requirements as control structures that restrict system behaviors at meta-model level. We propose the formalism of interface C-Systems, short for "interface control systems''. In this framework, functional requirements and safety requirements are separately formalized as interface automata and controlling automata respectively, as what we are doing in practice. The controlling automaton may guarantee safety requirements at design-time or runtime. Then the global system is a safe specification. The underlying mechanism differs from that of model checking. It explicitly separates the tasks of product engineers and safety engineers, and provides a new top-down methodology for designing and modeling a system with safety constraints, and for automatically composing a safe system that conforms to safety requirements. In practice, this methodology may be also used for safety checking, incident reporting and service restoration.