Profiling-Based Indoor Localization Schemes

In this paper, we consider the indoor localization problem, i.e., identifying the Cartesian coordinates of an object or a person under the roof. To solve this problem we consider an RF-based localization method called profiling, a two-step process, where a radio map of the monitored area is first constructed by collecting signal strength from known locations. An unknown location is then predicted using this radio map as a reference. In this paper, we first propose a K nearest neighbor (KNN) profiling-based localization method dubbed LEMON (location estimation by mining oversampled neighborhoods). It is based on a low-cost, low-power wireless devices and ensures good accuracy compared to the state-of-the-art. We then propose a variant of LEMON called combinatorial localization which exhaustively searches for the best possible set of nearest neighbors. We further define a Bayesian network model for the same localization problem. The performance of these methods is evaluated through extensive experiments in various indoor areas. We found an interesting outcome that the simple KNN-based approach can offer better localization accuracy compared to other complex localization methods. Thus we further enhance the performance of the KNN-based approach using multiple RF channels.