Fingerprint Similarity Indoor Localization: An AP Selection Based Approach

Abstract

With the flexible collection of received signal strengths (RSSs) by wide and ubiquitous indoor Wi-Fi infrastructures, the indoor fingerprint-based localization systems based on RSSs are gaining popularity. However, the device heterogeneity significantly degrades the performance of the fingerprint-based localization. Since different devices have different reception sensitivity, the RSSs collected by different devices at the same location for the same access point (AP) may be significantly different. To handle the device heterogeneity problem, we propose an AP selection based calibration-free method, which includes offline phase and online phase. In particular, an AP selection method is designed in the offline phase and selects the APs whose RSSs are of great difference for constructing fingerprint database. Moreover, a signal strength difference ranking method is designed in the online phase, which alleviates the impact of heterogeneous devices. Especially, experiments results validate that the proposed method could improve the region localization accuracy up to 23%, compared with the existing calibration-free methods.