A case-based approach for indoor location

Abstract

Location is an important dimension for context-awareness in ubiquitous devices. Nowadays different techniques are used alone or together to determine the position of a person or object. One aspect of the problem concerns to indoor location. Various authors propose the analysis of Radio Frequency (RF) footprints. In this paper we defend that case-based reasoning can make an important contribution for location from RF footprints. We apply an empirical dissimilarity metric for footprint retrieval and compare this approach with the results obtained with a neural network and C5.0 learning algorithms. The RF footprints are obtained from a Global System for Mobile Communications and General Packet Radio Service (GSM/GPRS) network. Signals from these networks are particularly complex when compared to the ones obtained from WiFi or Bluetooth networks. © Springer-Verlag Berlin Heidelberg 2005.