Research On Indoor Localization Algorithm In WLAN

With the development of computer science and application of the Internet,mobile localization technology has drawn extensive attention. Especially in the architecture of Io T and the platform of "Internet plus", the information of location,whatever outdoor or indoor, is indispensable. Traditional positioning technology based on satellites, such as GPS, can well meet the needs of outdoor location. But if we use the technology in the situation of indoor, the accuracy is very low because of blocking of satellite signals by buildings, and the high cost of implementation of this method, it is not suitable for the indoor positioning.Because of simple and wide deployment, low price and other outstanding features, WLAN is more suitable for indoor localization. Recently, researches on indoor positioning based on WLAN have become a hot topic in the field of indoor positioning. And the positioning method based on the architecture of the location fingerprints has become the dominant algorithm.In this paper, I have analyzed the implementation of the indoor localization algorithm based on the architecture of location fingerprints in WLAN environments.Through analysis, I pointed out the error during positioning and the shortcomings of several existed algorithms. Through deep analysis, a novel indoor localization algorithm based on KPCA and IWKNN is presented in this paper, called KPCA-IWKNN. The algorithm has the following improvements :1.The algorithm apply KPCA to train the original location fingerprints in the offline phase, KPCA can extract the nonlinear characteristics of the original location fingerprints. And we use these nonlinear characteristic to build feature-position-fingerprint database. Therefore we can effectively use the RSS signal of each AP.2.In the online phase, apply KPCA to train the online measured position fingerprint after, then using the proposed IWKNN algorithm. By setting a threshold parameter, the algorithm can decide the number of neighbor position fingerprints, and output the weighted positioning results. The proposed algorithm can avoid the problems caused by the stable K neighbors of the original KNN algorithm. And the proposed algorithm can improve positioning accuracy.3.The fingerprint data are derived from the real WLAN environment, so that the performance can be more close to the real WLAN environment. Then analyze the influence of the number of AP and RSS, etc. And compare the proposed algorithm with other algorithms.The simulation results show that the proposed algorithm is better than the other indoor location algorithm on positioning accuracy and the mean error. In the same location accuracy, the proposed algorithm requires less training samples and AP nodes,thus reduces the training consumption in the offline phase. And the proposed algorithm used in another real WLAN positioning environment, and compare with other algorithms, the results show that the proposed algorithm has stable performance.