University of Houston

The wireless localization problem is defined as determination of an assignment of coordinates to nodes in a wireless network which is consistent with the actual physical location of the nodes. In recent literature, several techniques addressing this problem under different context exists. However, even the best of the techniques require extensive site surveys for reasonable accuracy in addition to the need for special hardwares - making it less attractive for practical purposes.

In this thesis, we have focused on addressing this issue in the localization problem i.e to reduce the requirement for extensive site surveys while retaining the accuracy as much as possible. We have proposed a localization system based on surface approximation technique. Our approach utilizes the principle of locality for wireless signal strength (i.e. RSSI) of nodes that are in close geographical proximity by applying surface approximation algorithm to construct the signal strength map of stationary nodes over the region of interest and uses this approximated RSSI map to match the observed signal strength to infer location. In the thesis, we have evaluated our proposed method on real measurement data sets collected from indoor deployment of 802.11 networks and have shown the robustness of the method compared to the baselines.