Generating user interfaces for ambient intelligence systems

User Experience has become a key success factor for today's software applications. As the only part of a software application a user has direct contact with, the user interface contributes to a large degree to the overall user experience. This applies especially to ambient intelligence (AmI) systems that are more user centered than traditional systems. Although a system's software architecture is considered as one of the most significant technical factors in ensuring project success, its influence on user experience is usually neglected in today's software development. Thereby, the connection between the user interface part of a system and the system's core functionality is of particular importance. A system's distribution architecture and in particular the used client type has a significant influence on the overall user experience. The central contribution of this thesis is the introduction of an application's client type as an explicit adaptation factor in software development in general and in user interface generation in particular to improve user experience. As AmI systems are characterized by openness, heterogeneity and run-time dynamics, their user interfaces have to be generated to provide always an adequate user experience for the frequently changing usage situation. Thus, the appropriate client type needs to be determined automatically during user interface generation. To support developers in the client type selection, we provide a comprehensive characterization of nowadays commonly used client types: fat clients, thin clients, smart clients, rich internet clients and ultra-thin clients. Furthermore, we propose a method that allows the selection of an adequate client type for the current usage situation. Based on a detailed quality model that allows the assessment of the influence of several client type characteristics on the overall user experience and a usage situation model that describes context factors affecting the client type selection, the method determines the recommended client type automatically using a weighting function. With a concrete implementation of our client type selection method, supporting a fully automatic user interface generation approach that we applied within the ambient assisted living (AAL) domain, we demonstrate the feasibility of our approach. Finally, we describe further tool-support for user interface developers using a semi-automatic UI generation approach as well as for developers using a traditional software development process. Ultimately, the client type selection should be considered explicitly for every application during development.