Hybrid technique for fingerprinting using IEEE802.11 wireless networks

Location based on fingerprinting comprises two distinct phases: the first phase, off-line phase, is related with two tasks, a site survey to collect data for the Fingerprint Map (FM), and its generation based on the collected data; the second phase, on-line phase, is related with the location of a mobile terminal, by doing the comparison of the data acquired from the wireless transceiver with the data stored in FM. Typically the first phase is very time consuming because multiple samples of the wireless signals, received from the multiple references, must be acquired and stored, for every spatial point to be considered in the FM. In this paper a hybrid approach using propagation models for indoor environments and fingerprinting is presented. In this approach the FM is generated using indoor propagation models and the map of the scenario, instead of requiring a site survey. This makes the generation of the FM faster when compared with the traditional approach. Also when changes occur the FM can be easily recalculated. Several Location Estimation Algorithms were used to test the FM. A precision of 3.118m was achieved using k-Nearest Neighbour and a standard deviation of 0.133m was obtained using Fuzzy Logic.