A calculus for dynamic architectures

Abstract

• A set of 35 rules to reason about component behavior in a dynamic context is provided. • The rules are implemented in Isabelle/HOL and their soundness is verified mechanically in 3500 lines of Isabelle/HOL code. • The approach is evaluated by means of four case studies resulting in 3800 additional lines of Isabelle/HOL code. With the emergence of mobile and adaptive computing, dynamic architectures have become increasingly important. In such architectures, components may appear or disappear, and connections between them may change over time. Dynamic architectures are usually specified in terms of two separate specifications: a specification of component behavior and a specification of component activation and reconfiguration. To verify them, both specifications are first interpreted over a common model for dynamic architectures and verified against a property specified over the same model. Interpreting the specifications over the model, however, introduce repetitive proof steps, which increase the length of proofs, and thus the effort to develop and maintain them. To address this problem, we developed a calculus for dynamic architectures providing rules to reason about component behavior in a dynamic environment. We proved soundness and relative completeness of the rules, implemented them in the interactive theorem prover Isabelle, and mechanized the corresponding soundness proofs. The calculus can be used to support the abstract verification of dynamic architectures in Isabelle. This is demonstrated by means of a running example and evaluated in terms of four case studies. Our results suggest that the calculus has the potential to reduce the length of proofs for the verification of dynamic architectures, thus reducing the effort to develop and maintain verification results.