Extending Formal Methods for Software-Intensive Systems

Formal methods have proven beneficial in the industrial development of software-intensive systems; not in replacing traditional engineering methods, but in complementing them. They provide means of checking for ambiguities and inconsistencies in requirements, as well as verifying safety and liveness properties, and the correctness of designs. As complexity increases, the formal methods employed need to deal with a number of concerns. Primarily they need to be able to model a diverse range of software and hardware components. Ideally, they should also be capable of supporting requirement changes allowing `ideal' functional specifications to be transformed to reflect actual implementations. Additionally, they should support the introduction of architectural design into functional specifications; including designs involving complex dynamic architectures. This paper proposes one approach to deal with these concerns. The approach builds on and combines three separate areas of research on integrating formal methods, formal requirements development and formal design derivation. Developing more general theories and techniques that can be applied across a wide range of formal notations remains a significant research challenge.