802.11 Positioning in the Home

802.11 positioning systems are established as a low-cost solution to positioning within context aware computing. Prior research has mostly focused on either indoor positioning within commercial environments, where access points are prevalent, or outdoors positioning over large areas, using discovered networks to augment GPS. In this paper we test 802.11 positioning in a medium sized domestic house with only a small number of access points, examining its suitability in a domestic context aware computing system. We use a signal strength fingerprint map approach, in which empirical signal strength data is gathered over the area prior to use. Estimation then involves comparing the online input to the map, and selecting the best position, and direction, based on similarity. We compare different strategies for gathering signal strength, and implementations of the Nearest Neighbor and Bayesian methods. Our results demonstrate that with good placement, only two access points are sufficient to estimate position with an error of less than 4 meters 90% of the time.