A Probabilistic Clustering-Based Indoor Location Determination System

Abstract

We present an indoor location determination system based on signal
\nstrength probability distributions for tackling the noisy wireless
\nchannel and clustering to reduce computation requirements. We provide
\ntwo implementation techniques, namely, Joint Clustering and Incremental
\nTriangulation and describe their tradeoffs in terms of location
\ndetermination accuracy and computation requirement. Both techniques have
\nbeen incorporated in two implemented context-aware systems: User
\nPositioning System and the Rover System, both running on Compaq iPAQ
\nPocket PC's with Familiar distribution of Linux for PDA's. The results
\nobtained show that both techniques give the user location with over 90%
\naccuracy to within 7 feet with very low computation requirements, hence
\nenabling a set of context-aware applications.
\nAlso UMIACS-TR-2002-30