Family-oriented requirements engineering
Detlef Streitferdt · Common Library Network (Der Gemeinsame Bibliotheksverbund) · 2004
Abstract Modern software products shall be developed within a short time and at the same time they should be of a high quality. Software engineering is able to fulfill these requirements by prefabricating components. Within a domain, planned and comprehensive reuse of components is supported by the concept of system family development. A system family is based on a reference architecture made of assets, which are common to all family members and assets, which are variable. Commonalities and variabilities have to be considered in all phases of the system family development and they have to be elaborated correctly. Mistakes made in the requirements engineering phase, as the beginning of a development, will cause the most damage to the overall development. The analysis of the requirements engineering phase of system family development in this paper shows, that most of the current scientific solutions in this field are based on feature modeling, although it is inconsistent and cannot be automatically processed. The proposed solution of this paper – Family-Oriented Requirements Engineering (FORE) – extends feature modeling and integrates it into a new data model, capable of holding all the information acquired within the requirements engineering phase. Dependencies within feature models and between features and further model elements can be modeled with the new FORE Feature Constraint Language (FCL). FCL offers 30 predefined dependencies for verifiable system family models. Extended feature models of FORE allow the automated verification of a subset of features. Thus, only valid family members can be derived of the system family. The development of a system family model as well as the derivation of family members are supported by the FORE development process. The FORE-process integrates current processes and the extended feature modeling of FORE into a requirements engineering process for system families. The proposed solution was prototypically realized by implementing the FORE-Data Model as XML-Schema. The usage of the prototype is aimed at the FORE-Development Process. FORE was tested within a University project as well as in several student works with industry partners. The results of this paper have shown the applicability of FORE, improved its process and data model and revealed its limits.