Software architecture recovery and restructuring through clustering techniques

Abstract

1. ABSTRACT Capturing of software architecture is critical for maintenance and evolution. However, existing approaches often are limited only to software architecture recovery in the reverse engineering process. What is needed more is a systematic and effective approach to help the designer to restructure or reengineer an architecture for improvement. This paper presents an quantitative approach based on clustering techniques for software architecture restructuring and reengineering as well as for software architecture recovery. Clustering techniques are built on top of reverse engineering tools. The approach has been applied to several examples at various levels of abstraction. Two case studies are presented in this paper. One is an empirical study of a decoupling effort for a real-time telecommunications system. The other example shows a research potential to enforce the designer to improve an architecture by adopting a design pattern based on the clustering results. 1.1 Keywords Architecture recovery, restructuring, evolution, clustering, patterns