Enhancing a user context by real-time clustering mobile trajectories

Abstract

With the increasing availability of mobile networks and mobile devices, new paradigms of interaction between users and between a user and the surrounding environment are emerging. Location-aware systems [1-2] play a particular role in this new environments and, when the position of a user is known, applications running on the mobile devices, and even the virtual environment being visited by a user, can adapt their behaviour accordingly to the position of the user. In these systems, the position information is obtained from a set of position sensors, or from network-based location services, and is often used directly, without any further processing, as a parameter in the adaptation process. However, in addition to a low-level representation of positions as provided by position sensors, location-aware systems could benefit from having high-level representations of locations and places. In this paper we show that the context of a user can be enhanced with high-level location data useful for the adaptation of location-aware applications. For this, we propose an inference method based on a real-time clustering algorithm to automatically estimate the location of the places where a user lives and works. The achieved results show that the proposed approach rapidly converges to the expected locations, and that the same algorithm can be used to estimate both the home and workplace simultaneously.