Analysis of AP's Discrimination Capability for Fingerprint-Based Indoor Localization Technology

Fingerprint-based indoor localization has attracted more attention due to its simplicity, low cost, and compatibility. However, the RSS values of a position pair are collected from different APs with different statuses. Hence, the contribution of the RSS difference from different APs to the discrimination capability of fingerprint distances is different. In this paper, we recognized APs' diverse discrimination capability for different pairwise positions from AP's health state, the distance, and the direction between the AP and the position pair. Inspired by these insights, for a position pair, we devise a discrimination weight based on predefining the impact factor of AP and a priority token to identify the priority of the RSS difference of AP participating in the fingerprint distance and eliminate abnormal APs by means+3S. Integrating these perspectives, some policies for calculating the fingerprint distance for each position pair are proposed. During the simulation, we first study the effects of overall APs distribution on the correlation and consistency between fingerprint distances vector and physical distances vector of all pairwise positions and point out that the optimal position combination of APs is evenly distributed at the edge of the localization area. Also, we evaluate the policies, the initial results show that the correlation and consistency between fingerprint distances and physical distances of the pairwise position are improved to some extent, demonstrating that the algorithm has strong robustness to abnormal APs.