The feature architecture mapping method for feature oriented development of software product lines

Software product lines are the answer of software engineering to the increasing complexity and shorter time-to-market of contemporary software systems. Nonetheless, software product lines demand for advanced maintainability and high flexibility. The latter can be achieved through the proper separation of concerns. Features pose the main concerns in the context of software product lines. Consequently, one feature should ideally be implemented into exactly one architectural component. In practice, this is not always feasible. Therefore, at least a strong mapping between features and the architecture must exist. The state of the art product line development methodologies introduce significant scattering and tangling of features. In this work, the Feature-Architecture Mapping (FArM) method is developed, to provide a stronger mapping between features and the product line architecture. FArM receives as input an initial feature model created by a domain analysis method. The initial feature model undergoes a series of transformations. The transformations strive to achieve a balance between the customer and architectural perspectives. Feature interaction is explicitly optimized during the feature model transformations. For each feature of the transformed feature model, one architectural component is derived. The architectural components implement the application logic of the respective features. The component communication reflects the feature interaction. This approach, compared to the state of the art product line methodologies, allows a stronger feature-architecture mapping and for higher variability on the feature level. These attributes provide higher maintainability and an improved generative approach to product instantiation, which in turn enhances product line flexibility. FArM has been evaluated through its application in a number of domains, e.g in the mobile phone domain and the Integrated Development Environment (IDE) domain. This work will present FArM on the basis of a case study in the domain of artificial Neural Networks.