Qualitative Positioning for Pervasive Environments

In this paper we present a strategy and a set of algorithms for developing qualitative positioning services that provide a qualitative location optimised for the environment where they are to be deployed. We argue that for many contextaware applications this may be more appropriate than more common quantitative location systems, where the positioning API may make unrealistic demands on the underlying measurement service, and unrealistic promises to the application. We show how a symbolic location system can be learnt from training data in an unsupervised manner. We present experimental results using 802.11 and GSM signal strength levels and wireless beacon data.