Supporting Software Architecture Evolution by Functional Decomposition

Software systems evolve during their lifetime to reflect the changes in their users needs. However, unless implemented carefully, such changes may degrade the quality of the system’s architecture by reducing the cohesion and increasing the coupling between its subsystems. It is therefore important to systematically analyze the changes and modify the system’s structure to accommodate the changes without degrading the system’s architecture. However, looking just at functional aspects is not enough, because we may decide on a redesign that is too expensive to implement. In this paper we combine a functional decomposition analysis technique with a nonfunctional impact analysis technique to avoid this pitfall. The functional decomposition technique generates a set of plausible decompositions that accommodate the required evolutionary changes, and the impact analysis technique acts as a filter that selects only those decompositions that satisfy the cost constraints of the required changes. We briefly describe both techniques and then illustrate the approach with an example of a parking lot management system.