Εντοπισμός θέσης και ιχνηλάτηση μονοπατιού με χρήση της υποδομής ΙΕΕΕ 802.11

The deployment of mobile networks is combined with special services such as geolocalization inside public buildings. This geolocalization cannot be established by the classic ways of GPS, which does not work indoors. The geolocalization from GSM triangulation is not precise enough. This thesis propose the development of a geolocalization’s protocol based on the mesh structure of the network and the application of the statistic methods known as fuzzy logic and particle filters. It is an improvement of the Kalman Filter method, without any constraint or assumption about the node mobility. The localization algorithm is based on an analytical drift evaluation made with signal strength measure- ments of the 802.11 infrastructure. At each step, the localization algorithm deduces the optimal correction in the mobile node’s position based on the current estimated location, the signal measurements and the known signal distribution database. In fact, the algorithm relies on an analytical evaluation of the perturbation of the signal strength distribution with respect to the most likely path, which corresponds to the discrepancy between the real and estimated positions. At each step, the algorithm aims to correct this drift in the node position. Consequently, the corrections performed result in a random walk, which on average follows the real path taken by the mobile node, i.e. the expectation of the error in the estimated position is 0. The performance of our approach is evaluated by numerical simulations, as well as, by using real measurements