Options
2006
Doctoral Thesis
Titel
Towards an integrated approach for software process improvement: Combining software process assessment and software process modeling
Abstract
Software-Prozessverbesserung (SPV) wurde in den letzten Jahren immer wichtiger, da der Wettbewerb zunehmend durch die Softwareanteile an Produkten und Dienstleistungen entschieden wird. Der Einsatz von etablierten Praktiken in der Softwareentwicklung verspricht einen Wettbewerbsvorteil und eine höhere Kundenzufriedenheit. In den letzten Jahren wurden viele verschiedene SPV-Ansätze entwickelt. Diese Dissertation stellt einen integrierten SPV-Ansatz vor, der ein Prozessassessment dahingehend erweitert, dass die Informationen bzgl. Produkten und ihren Beziehungen zu den im Assessment untersuchten Prozessen schon während dem Assessment gemäß einem konzeptuellen Modell gesammelt und dokumentiert werden. Dadurch kann ein in sich konsistentes Prozessmodellskelett automatisch generiert werden, das als Basis für die weiteren Prozessmodellierungsaktivitäten dient, die ebenfalls entsprechend angepasst wurden. Die Validierung durch ein Experiment und eine industrielle Fallstudie zeigt, dass die Integration von Prozessassessment und Prozessmodellierung sowohl die Qualität des Assessments als auch die des entstehenden Prozessmodells erhöht, bei gleichzeitiger Verringerung des SPV-Gesamtaufwandes.
;
Software Process Improvement (SPI) has become more and more important during the past few years, since competition is increasingly determined by the software portion of products and services. The use of best practices in software development promises to provide a competitive advantage and to achieve higher customer satisfaction. Over the years, many different SPI approaches have been developed. Most of them are either best-practice-oriented approaches (e.g., CMM(I), SPICE), or continuous improvement approaches (e.g., PDCA, QIP, PROFES), which require an accurate process description as their basis. Almost always companies choose only one of them to start any improvement activity, and continue with another technology separately later on, which leads to the artificial separation of SPI activities and unnecessary effort consumption. The reason is that these SPI activities are not consistent as such nor comparable to each other. The main achievement of this research lies in the development and evaluation of an integrated methodology for assessing and modeling processes based on the PROFES framework and descriptive modeling concepts. The contributions are: a) an overview of the state-of-the-art and state-of-the-practice for SPI approaches, b) a set of requirements for an integrated approach for SPI, c) a formal conceptual model that integrates software process assessment and software process modeling concepts, d) a process description of the integrated SPI methodology including practical guidelines, e) validation in an experimental setting by means of a quasi-experiment followed by a case study in an industrial company, and f) tool support on the basis of the conceptual model. These contributions have been incorporated into the Fraunhofer Assessment Method (called FAME). As part of FAME, the assessment phase has been enhanced in such a way that the information about products and their relationship to the processes under evaluation in the assessment is explicitly elicited and documented according to the conceptual model. This allows the automatic generation of a consistent process model skeleton that can be used as the basis for all subsequent process modeling activities, which have consequently also been adopted accordingly. By considering the requirements of the upcoming revision of the process assessment standard ISO/IEC IS 15504, this forms a comprehensive SPI method including tool support, which is ready for roll-out in industrial practice. The benefits of the integrated methodology have been demonstrated by an empirical validation. The results of a quasi-experiment show that a FAME assessment produces more accurate process assessment ratings (~12% improvement), including a higher amount of documented process information (increase by ~8%), than a traditional SPICE assessment. Additionally, the results of the industrial case study show that the integrated methodology produces higher quality process models (~50% less inconsistencies in the first iteration) with less effort (~10%) than a traditional modeling approach. Altogether, it can be shown that the integration of process assessment and process modeling raises the quality of the assessment as well as the resulting process model, including effort reduction in the course of software process improvement.
ThesisNote
Zugl.: Kaiserslautern, TU, Diss., 2006
Language
English