Measuring the quality of transformation alternatives in software architectures evolution

Abstract

Many today's software systems need to be self-adapted at run-time. Model transformation is a good approach to adapt the component-based architecture of such software systems. However, existing model transformation processes focus on the functionalities of systems, giving less importance to the quality attributes. The goal of this study is to improve model transformation processes by also considering quality attributes in the generation and adaptation of component-based architectures (i.e., driving the selection among many alternative model transformations by software architecture metrics). Such metrics evaluate the qualities of an architecture, such as flexibility and modifiability. This paper provides some measures of quality for different transformation alternatives and an example in the ENIA software.