AP-sequence Based Fingerprint Similarity Indoor Localization

Due to the extensive deployment of WiFi access points (APs), WiFi based localization methods have such tremendous potential and have been extensively applied for indoor localization. However, if not properly handled, two critical issues including massive radio map construction overhead and the timevarying feature of fingerprints, e.g., received signal strength (RSS), would result in significant performance degradation of the available indoor localization methods. To overcome the bottleneck, an AP-sequence based fingerprint similarity indoor localization approach (APFS) has been designed in this work. In particular, to achieve extremely low overhead in the offline phase, the ordered AP sequences are applied to construct the fingerprint in the radio map of APFS. Specifically, the ordered AP sequences are obtained by sorting the distances from the reference points (RPs) to the available APs in the ascending order. Moreover, to handle the issue caused by time-varying RSSs, an algorithm based on fingerprints similarity is designed in the online phase. Specifically, the algorithm could tolerate AP sequence disorders caused by time-varying wireless channels. Meanwhile, the APs that have stronger RSSs would be assigned with higher weights, since they would contribute more during the matching of the measured fingerprint and the radio map. In addition, the relative RSS values rather than the absolute RSS values are adopted as the online fingerprints for practical concerns. In consequence, the RSSs inaccuracy at test points (TPs), which is caused by the RSSs collection by different devices, could be avoided. Experimental results have been presented to show that APFS could achieve better performance in terms of localization accuracy, compared to the conventional selective AP-sequence method under typical system setups.