Using Weighted Graphs for Computationally Efficient WLAN Location Determination

Indoor location-based services hold promise for a multitude of valuable services, but require micro-detailed geo-referencing not achievable with "outdoor" technologies such as GPS and cellular networks. A widely used technique for accurate indoor positioning is location fingerprinting which makes use of existing WLAN infrastructures. The technique consists of building a radio map of signal strength measurements which is searched to determine a position estimate. While the fingerprinting technique has produced good positioning accuracy results, the technique incurs a substantial computational burden for large buildings and is thus problematic for tracking users in real time on processor-constrained mobile devices. In this paper we present a technique for improving the computational efficiency of the fingerprinting technique such that location determination becomes tractable on a mobile device. The technique is based on a graph-modeling of the physical environment and works by restricting the search space to positions that are possible to reach from a previously estimated position. The technique is general in that it can be applied in conjunction with any positioning algorithm, and a positive side effect is that it may enhance the positioning accuracy of the system