Dokument

Summary:
Model-driven development (MDD) of software systems has been a serious trend in different application domains over the last 15 years. While technologies, platforms, and architectural paradigms have changed several times since model-driven development processes were first introduced, their applicability and usefulness are discussed every time a new technological trend appears. Looking at the rapid market penetration of smartphones, software engineers are curious about how model-driven development technologies can deal with this novel and emergent domain of software engineering (SE). Indeed, software engineering of mobile applications provides many challenges that model-driven development can address. Model-driven development uses a platform independent model as a crucial artifact. Such a model usually follows a domain-specific modeling language and separates the business concerns from the technical concerns. These platform-independent models can be reused for generating native program code for several mobile software platforms. However, a major drawback of model-driven development is that infrastructure developers must provide a fairly sophisticated model-driven development infrastructure before mobile application developers can create mobile applications in a model-driven way. Hence, the first part of this thesis deals with designing a model-driven development infrastructure for mobile applications. We will follow a rigorous design process comprising a domain analysis, the design of a domain-specific modeling language, and the development of the corresponding model editors. To ensure that the code generators produce high-quality application code and the resulting mobile applications follow a proper architectural design, we will analyze several representative reference applications beforehand. Thus, the reader will get an insight into both the features of mobile applications and the steps that are required to design and implement a model-driven development infrastructure. As a result of the domain analysis and the analysis of the reference applications, we identified context-awareness as a further important feature of mobile applications. Current software engineering tools do not sufficiently support designing and implementing of context-aware mobile applications. Although these tools (e.g., middleware approaches) support the definition and the collection of contextual information, the adaptation of the mobile application must often be implemented by hand at a low abstraction level by the mobile application developers. Thus, the second part of this thesis demonstrates how context-aware mobile applications can be designed more easily by using a model-driven development approach. Techniques such as model transformation and model interpretation are used to adapt mobile applications to different contexts at design time or runtime. Moreover, model analysis and model-based simulation help mobile application developers to evaluate a designed mobile application (i.e., app model) prior to its generation and deployment with respected to certain contexts. We demonstrate the usefulness and applicability of the model-driven development infrastructure we developed by seven case examples. These showcases demonstrate the designing of mobile applications in different domains. We demonstrate the scalability of our model-driven development infrastructure with several performance tests, focusing on the generation time of mobile applications, as well as their runtime performance. Moreover, the usability was successfully evaluated during several hands-on training sessions by real mobile application developers with different skill levels.

Das Dokument ist im Internet frei zugänglich - Hinweise zu den Nutzungsrechten