AP weighted multiple matching nearest neighbors approach for fingerprint-based indoor localization

With the development of information technology, people's demand for location-based services is gradually increasing. Indoor localization and navigation is becoming increasingly popular. Wi-Fi fingerprinting localization is a low-cost and feasible method. Due to the impact of the AP (Access Point) itself and the indoor environment, real-time AP signal strength value will be volatile. In order to keep the AP signal features more complete, this paper presents a new matching approach for localization. In the offline stage, repeatedly acquired signal values are saved to the fingerprint database. In the online stage, the APs of test and reference points with multiple signal strength values were compared, record the absolute value of the minimum difference as the difference of reference point and testing point of the AP, followed by matching each AP and the weight center position of K reference points with minimum differences as the final coordinates of unknown point. At the same time, the weight of the signal is added in the calculation process, which is used to distinguish the AP of different distance. Experimental results show that the proposed approach can improve the accuracy of localization obviously compared with the traditional WKNN algorithm and Gaussian Kernel algorithm.