Automated semantics-preserving parallel decomposition of finite component and connector architectures

Abstract

For the systematic development of logical, message-driven architectures, automating parallel decomposition of software components is important to achieve efficient modular and parallel system development. During development, monolithic components that realize multiple independent concerns need to be decomposed to obtain a higher quality architecture of cohesively encapsulated, better comprehensive components. Previous work did not address automated parallel decomposition of finite message-driven and logically timed components with respect to the influence of messages received via input channels on the messages sent via output channels. This, however, is a necessary prerequisite to enable the analysis of event chains across logically distributed architectures. To address this, we present a concept of influence between channels of components that supports automated semantics-preserving parallel decomposition of finite deterministic component implementations into independent, more comprehensible components that are better accessible for analysis and development. Therefore, we extend the Focus theory of time-synchronous components with the concept of influence, present a decomposition procedure leveraging this, and prove that the resulting system is semantically equivalent. This enables automatically decomposing monolithic software components (e.g., for stepwise refinement or refactoring) into smaller components of better cohesion and comprehensibility and thus facilitates automated software engineering.