Is software aging related to software metrics?

This work presents an empirical analysis aiming at investigating what kind of relationship exists between software aging and several static features of the software. While past studies on software aging focused on predicting the aging effects by monitoring and analytically modeling resource consumption at runtime, this study intends to explore if the static features of the software, as derived by its source code, presents potential relationships with software aging. We adopt a set of common software metrics concerning program structure, such as size and cyclomatic complexity, along with some features specifically developed for this study; metrics were then computed from ten complex software applications affected by aging. A statistical analysis to infer their relationship with software aging was carried out. Results encourage further investigations in this direction, since they show that software aging effects are related to the static features of software.