Online adaptive positioning algorithm for public location services in indoor places

WLAN has become popular in recent years because of the universality and economy. It has a large prospect of the application on the positioning technology in indoor places. However it is difficult for the traditional indoor positioning method based on WIFI to adapt to the changing environment in public places. In order to remedy the defect, this paper proposed a real-time online adaptive fingerprint positioning algorithm. The probabilistic method of Bayesian estimation is applied to the underlying algorithm. Then the positioning process proposed in this paper will include the real-time reduction of the fingerprint vectors ´dimension based on Fisher criterion and besides, and dynamically filtering out the probably grids by the K-means theory during the online phase. Hence it makes possible to adjust the number of grids and their corresponding fingerprint vectors dynamically which will be used for the calculation of probability later. Field tests showed that, compared with the traditional nearest neighbor method, the average distance error of the algorithm was smaller while still maintaining acceptable efficiency of computation to meet public traveler´ needs.