Building a personal symbolic space model from GSM CellID Positioning Data

Abstract

The context in which a person uses a mobile context-aware application can be described by many dimensions, including the, most popular, location and position. Some of the data used to describe these dimensions can be acquired directly from sensors or computed by reasoning algorithms. In this paper we propose to contextualize the mobile user of context-aware applications by describing his/her location in a symbolic space model as an alternative to the use of a position represented by a pair of coordinates in a geometric absolute referential. By exploiting the ubiquity of GSM networks, we describe a method to progressively create this symbolic and personal space model, and propose an approach to compute the level of familiarity a person has with each of the identified places. The validity of the developed model is evaluated by comparing the identified places and the computed values for the familiarity index with a ground truth represented by GPS data and the detailed agenda of a few persons.