Virtual WiFi network database construction for positioning services

WiFi-based location based services (LBSs) are expected to be more and more pervasive in future Internet. In WiFi networks, a virtual network database (VNDB), which maintains the estimated locations of access points (APs), should be constructed for the localization. Since the quality of the LBS is mainly determined by the location accuracy, it is essential to maintain accurate VNDBs for the LBS. In this paper, we propose two novel filtering methods to construct accurate VNDBs for the WiFi-based localization: RSS filtering and grid-based filtering. Experimental results on a real testbed show that the proposed methods construct a VNDB accurately and localization results with the accurate VNDB are comparable to those with a real network database (RNDB).