Indoor positioning is the hard demand of smart cities and many smart city-related applications are inseparable from location services. At present, the main indoor positioning technology includes Bluetooth, RFID, UWB and geomagnetic, etc., but due to such issues as cost, deployment convenience and so on, its applications development are limited. This paper proposes a KNN(K-Neatest Neighbor) localization algorithm based on fingerprint timing features (TS-KNN), which uses the fingerprint of the current moment to select the reference coordinates and the positioning results of the first few moments are used to perform weight correction for each reference coordinate. The experimental test results in a square in Chongqing show that the proposed TS-KNN method is superior to the KNN and the WKNN algorithms, since it can effectively improve the accuracy of indoor positioning and reduce the average positioning error.