A Wifi Positioning Method Based on Stack Auto Encoder

Abstract

Based on the traditional Wifi positioning algorithm, the fingerprint database has large redundant information, which is easy to cause dimensionality disaster. In view of this situation, this paper proposes a Wifi positioning method based on ReliefF-SAE. Firstly, the feature selection of the constructed Wifi fingerprint database is carried out, which can remove the redundant information existing in the fingerprint database. Then the reduced fingerprint database is established. The deep neural network is constructed by stacked auto encoder, and pre-training is carried out to obtain a more accurate Wifi indoor positioning model. The positioning experiment is carried out by using the standard database UJIIndoor Loc. The experimental results show that the algorithm can effectively remove the redundant information in the fingerprint database, and at the same time, through the deep neural network learning, better positioning accuracy can be obtained.