Feature Extraction in Wireless Personal and Local Area Networks

Context awareness is currently being investigated for applications in different application areas of mobile computing. The integration of Bluetooth and Wireless LAN technologies into a vast of mobile devices — ranging from smartphones and PDAs to portable computers – has made user context sensing based on those technologies a feasible and promising approach. In this paper, we study which of the Bluetooth and Wireless LAN technology features (like radio-signal strength, device address management, etc.) can be exploited to derive user context, and develop a procedure how low level sensor data can be brought to application level context information. We introduce a method to automatically classify heterogeneous sensor data features with supervised or un-supervised classification methods. By defining two operations, a distance metric and an adaptation operator, any feature can be used as input for the classifier and can thus contribute to context detection.