Abstract: Considering the accuracy of WI-FI is limited to multipath effect in complex environment, an adaptive indoor localization algorithm based on CFSFDP was proposed. The algorithm is divided into three stages: the first in the pretreatment stage, the CFSFDP method is used to train the original fingerprint, and mining out stable and effective fingerprint features; the second in the offline stage, the sampling point strategy of the multi-layer coverage is further constructed, and the establishment of fingerprint; The third online phase, the RSS signal is used for parameter training. An adaptive signal ranging model is established to combine distance measurement method and fingerprint-based scheme, in order to make up for the disadvantage of the ranging scheme, and the ranging scheme reduces matching cost of the fingerprints in the online stage. Simulation results show that the proposed algorithm has obvious advantages in complex environment, and the ranging model has better performance.