A DCA-based method for software prognostics and health management

In recent years, the prognostics and health management (PHM) technique has greatly been developed especially for electronic devices and systems. Little work, however, has been done on PHM techniques for software systems. During their total lifecycle, the software systems, due to various reasons such as new requirements or changes in technology, have to experience changes that can lead to a degeneration of the system´s architecture. Therefore, it is important and necessary to identify and predict the software health status in order to avoid serious degeneration or erosion of system´s architecture. In this paper, a software PHM approach based on Discriminant Coordinates Analysis (DCA) is proposed, and the health status of software systems is suggested to be categorized into five groups: Healthy, Normal, Sub-healthy, Unhealthy and Deteriorated. The proposed model based on software design metrics can help managers and developers predict and identify the software health status. Numerical example based on simulation data is given to illustrate its application on the modeling process.