Fingerprint-based location estimation with virtual access points

Abstract

Location fingerprinting techniques generally make use of existing wireless network infrastructure. Consequently, the positions of the access points (APs), which constitute an integral part of a location system, will invariably be dictated by the network administrator's convenience regarding data communication. But the localization accuracy of fingerprint-based solutions is largely dependent on the APs' placements over the area. In this paper, we developed the idea of Virtual Access Point (VAP), where one can have AP's functionality at a desired position for localization purpose, without physically placing an AP there. We argue that, placing VAPs at favorable positions helps to improve localization accuracy. VAP also serves the purpose of virtually increasing the number of APs over the localization area, which according to previous works should enhance the localization accuracy further. We test the feasibility of our VAP idea both analytically and experimentally. Finally, we present our results using a well-known localization algorithm, namely, K-Nearest Neighbor, when our VAP idea is implemented. The findings are quite encouraging, which report significant improvement in the localization accuracy. & 2008 IEEE.