Fundamental Services for Context-Sensitive Mobile Applications Stephan Kopf, Thomas King, Philipp Bostan, Hendrik Lemelson, Sina Deibert, Wolfgang Effelsberg University of Mannheim 68131 Mannheim, Mannheim, Germany {kopf | king | bostan | lemelson | deibert | effelsberg}@uni-mannheim.de ABSTRACT A challenge for the development of mobile applications is the heterogeneity of the different device classes. For instance, an application which runs well on a personal digital assistant (PDA) might be unsuitable for mobile phones. To facilitate the development of mobile applications we have developed several services. We will elaborate on three fundamental services which simplify the user interaction: service discovery, indoor positioning, and image adaptation. These services enable a user- centered configuration of mobile applications. The search for a special service based on text queries or web browsing may be long-winded, because many services are not suitable for certain mobile devices. We have developed a context-sensitive service discovery technique which analyses user preferences and user context to select suitable services. An important context attribute is the current position of a user. Therefore, besides outdoor positioning services, we have developed an indoor positioning service, which reliably detects the position of a mobile device in buildings. The visualization of multimedia data is another important aspect for the human computer interaction due to the different display resolutions of mobile devices. Exemplarily, we present the image adaptation service which identifies relevant semantic content in images and adapts an image in an intelligent way to the screen resolution of a mobile device. Categories and Subject Descriptors H.5.2 [Information Interfaces and Presentation]: User Interfaces – Graphical user interfaces (GUI), User-centered design. C.2.1 [Computer Communication Networks]: Network Architecture and Design – Wireless communication. General Terms Algorithms Keywords Context-sensitive service discovery, indoor positioning, image adaptation. 1. INTRODUCTION Although, a great number of people in developed countries use mobile devices, the number of applications which are available on all the different mobile device classes is still relatively low. Usually, applications are only developed and tested for a certain device class (e.g., PDAs) and have to be adapted for other device classes (e.g., smart-phones). The porting of applications to the wide variety of mobile device classes is a time consuming issue. Additionally, testing and maintainability of applications is getting cumbersome for such a huge number of different mobile device classes. Fundamental services that support the development of applications for those heterogeneous mobile devices classes and in general middleware to support context-sensitive applications are still rare. In contrast to typical desktop applications, mobile applications are different since they need to consider mobility, resource limitations, heterogeneity, personalization and especially different requirements on usability and usage patterns. Particularly personalization which is bound to the context of the user, his device and the current environment, is an important issue for mobile applications. Context information like the current position and the environment of a user may help to select more suitable services, information or simply to adapt a mobile application to the current context or situation. Typical examples for context- sensitive applications are service discovery services, which intend to deliver services or information to the user that best fit the current context. For instance, in case of a restaurant finder application, it would be helpful if the user’s current position, his preferences like ‘non-smoking’, ‘prefer Italian food’, ‘outdoor seats’, or ‘not too expensive’ are considered. If the calendar of a user indicates other appointments in the near future the proposed restaurants should not be too far away (depending on available transportation). In the case of context-sensitive service discovery, it is important to provide users with enhanced support for finding suitable services in an efficient way. Long-winded interactions between users and mobile devices (e. g., browsing a huge list of services) in the process of service discovery should be avoided since users are restricted in their input capabilities. A middleware, which enables efficient service discovery, is a solution for this problem. Context-sensitive service discovery delivers significant added value since it considerably reduces the required level of interaction and delivers personalized, precise search results that are suitable in the user’s current context. We have developed a context-sensitive service discovery approach which is presented in Section 2.1. Since Schilit et. al. initiated research on context-aware computing in 1993 starting with their PARCTab project [28], various