An effective preprocessing Scheme for WLAN-based fingerprint positioning systems

Abstract

In WLAN-based fingerprint positioning systems, the data preprocessing before applying the matching algorithm is important for improving the accuracy. In this paper, an effective preprocessing scheme is proposed to improve selection methods of reference point (RP) and access point (AP). The spatial filter is used to select RPs, and APs are decided in the principle of minimal redundancy. The experiment is carried out on campus, whose results show that the proposed scheme could improve the accuracy of both the Kernel-based and KNN matching algorithm, especially the former, by about 25%. Moreover, the denser is the AP distribution, the better the proposed scheme performs.