An improved WiFi fingerprint location method for indoor positioning

Following the growth of wireless communication technology, indoor positioning is increasingly evolving and the location fingerprint positioning based on WiFi has gradually become a hot spot in the field of positioning technology. However, interferences for instance multi-path, shadow and noise effect in indoor area will cause WiFi signal fluctuation, which leads to deviation of fingerprint data. WiFi signal attenuation with distance is nonlinear so the traditional signal distance cannot accurately represent the physical distance. In order to handle the aboves, this paper proposes an improved WiFi fingerprint location method. A hybrid filtering algorithm based on Grubbs test is designed to reduce the noise of sampled RSSI (Received Signal Strength Indication) in the offline part. A weighted Euclidean distance is designed to improve the WKNN in the online part. The results show that the proposed filtering algorithm eliminates the gross errors and accidental errors. The improved WKNN could observably reduce the location error. The localization accuracy can reach 1.87m, which is 17.3% better than 2.26m of KNN and 3.1% better than 1.93m of WKNN.