In recent years, with the increasing demand for indoor positioning, the indoor devic-less passive positioning technology based on WiFi channel state image information has developed rapidly. we propose TR-KWNN positioning algorithm. The positioning algorithm based on Time Reversal technology (TR) equalized the reverse channel to a matched filter, and used the spatio-temporal focusing characteristics of the signal to achieve high-precision position matching, so that the minimum average positioning error was 0.678m using all link information. The positioning error is below 0.78m with 80% probability, which realizes high-precision device-free positioning of static targets in indoor environment. At the same time, in order to extend the ability of positioning technology and realize the continuous positioning of indoor personnel, a variable-length continuous fingerprint separation method that is more in line with actual application scenarios is designed. It greatly reduces the cost of fingerprint database acquisition.This method introduces the voice signal endpoint detection algorithm into the WiFi positioning field by continuously sampling the target position information, and completes fingerprint separation by marking the movement time of personnel. The proposed method can effectively solve the complex problem of offline fingerprint database acquisition, and the average positioning accuracy of all the separated fingerprint data is 0.796m. Finally, the integrated system of indoor passive positioning and tracking with low complexity and continuous stability is realized.