Abstract

Context-aware computing is proposed as an enabling technology for adaptating different functions in computer-aided devices. The development of context-awareness for mobile devices requires recognition and extraction of implicit context information from the usage situations and environment of a device. Context information is provided for applications and services, which adapt their appearance and functions accordingly. Mobile devices contain several potential context data sources such as, location, time and applications. In this thesis, sensors integrated into a mobile device are utilised as sources for context information.

The main challenge in sensor-based context-aware computing for mobile devices is how to define and carry out context recognition from sensor signals to facilitate use of context information in mobile applications. In this thesis we have divided this into specific research problems: How should low-level context information be extracted from sensor signals to obtain a rich and usable representation? How should low-level context information be processed and examined to obtain higher-level contexts? How to utilise context representation in applications? An empiric and data centric approach including signal processing, feature extraction and explorative data analysis methods is used in examining and defining a procedure for sensor-based context recognition.

The main result of this work is a procedure for sensor-based context recognition that is demonstrated with experiments and with the applications developed. The main technical solutions developed include methods for extracting context information and converting it into a suitable context representation, solution for collaborative recognition of the context of a group of mobile devices, an approach for controlling mobile applications and a solution for enhancing remote communication with context information. The thesis includes a review of context data processing and the utilisation of context information in mobile devices.