A Goal-oriented Mixed-granularity Component Selection Method for Huge Component Repositories

In component-based software development, component selection is a critical step since the selected components considerably affect the system quality and the integration cost. Traditionally, the selection has been done in an ad-hoc manner, which takes a long time and does not guarantee the quality of the resultant system. These problems become more significant as either the complexity of a system or the size of component repositories increase. Although a number of selection methods have been proposed, they cannot ensure the suitability of selected components on system-level and fail to consider mixed-granularity components. We present a goal-oriented mixed-granularity (GOMG) component selection method to solve these problems. It adopts a hierarchical goal tree to provide options to use components in different granularities. Also, it exploits a hierarchical evaluation method that systematically measures the appropriateness of component sets on the system level. We have conducted a case study of building a composite SoC CAD tool. The component repository consisted of 250 components that varied considerably in both granularity and performance. The results show that our GOMG method yields a better aggregated score than the existing method by 33% and reduces the time consumed on the selection by 73%.