Identification and Selection of Components using Component Coupling and Interface Metrics

Abstract

Modern software industry is now looking forward to Component Based Software Engineering (CBSE) for maximizing the software reusability. So CBSE is becoming the modern trend for software development. CBSE is focused on assembling the existing components to build a software, with the potential benefits of reduced development cost, time and improved quality. It differentiates itself from the conventional software development approach by concentrating highly on component integration to build a software rather than developing the software from scratch each time. The quality of the Component Based Software (CBS) depends upon the complexity of composed components. Thus evaluation of component complexity becomes a critical activity during component selection in Component Based Software Development (CBSD). So many researchers have proposed various complexity metrics for measuring component complexity during components selection for CBS. But many of the existing complexity metrics are not suitable and sufficient for measuring component complexity in the unavailability of source code or internal details of components. In this thesis, two complexity metrics, one based on measuring component Interface Complexity (ICBB) and another based on measuring component Coupling Complexity (CCBB), have been proposed which do not depend on source code or internal details of component. These metrics have been empirically evaluated. Further, correlation study has been conducted for these metrics with the quality characteristics and other aspects. The metric ICBB has been correlated with quality characteristics named portability and reusability, and it has been found that high complexity of the component interface leads to less portability and reusability of component. The metric CCBB has been correlated with the effort for replacement or modification and testing effort. The positive relation between them shows that as the component coupling complexity increases effort for component replacement or modification, and testing effort also increases.