Mobile Habits: Inferring and predicting user activities with a location-aware smartphone

In this paper we present a work in progress dedicated to the recognition and prediction of mobile user activities. In contrast with related projects that generally use GPS for localization, we employ a fusion of wireless positioning methods available in current smartphones (GPS, GSM, Wi-Fi). Our positioning system offers high availability and accuracy without dedicated calibration. We demonstrate how such a positioning information can improve place extraction algorithms and enable the recognition of the new types of user activities both indoors and outdoors. Besides that, the project addresses a number of open challenges in activity and place prediction, such as detection of behaviour changes, prediction of unseen places.