Generalizing the Submodule Construction Techniques for Extended State Machine Models

In previous research we extended the submodule construction techniques to cover a more expressive and compact behavioral model that handles data through parameterized interactions, state variables, and simple transition guards. The model was based on extended Input/Output Automata, and the algorithm on the Chaos concept. In this paper we generalize these extensions and improve the submodule construction techniques and algorithms. The generalizations include regular transition guards including equality and inequality, negation, conjunction and disjunction of predicates. The algorithm is improved by utilizing the concept of generic transitions (non refined transitions) that are refined as needed instead of considering all possible refinements of the Chaos. The algorithm selects needed refinements through dataflow relations bridging which involves forward propagation of definitions and backward propagation of usages. The new approach provides a more intuitive explanation of the submodule construction algorithm, gives justification for the number of variables in the new module and results in a much smaller and compact solution.