WiFi Fingerprint Localization Algorithm Based On Deep Learning

In the Internet of Things era,there is a growing demand for location coordinates,especially for indoor scenarios.Traditional GPS positioning cannot be effective in complex indoor environments,and indoor positioning technology based on WiFi signal fingerprint has received extensive attention due to its low cost and simple algorithm.In this paper,the deep neural network algorithm in machine learning is combined with the fingerprint characteristics of WiFi signal to carry out the research on target location in complex indoor environment.On the basis of extensive review of relevant literature at home and abroad,the development status of indoor positioning technology and the research background of deep neural network are introduced.Aiming at the problem of fingerprint database construction,this paper selects the UJIIndoorLoc fingerprint database in the machine learning library of the University of California.Through the analysis of the fingerprint database,the relevant dimension information is extracted for algorithm verification.In the processing of fingerprint database data,the method of reducing the number of APs and filtering the fingerprint sample points is used to optimize the data size of the fingerprint database,reduce the noise interference caused by signal fluctuations,and improve the robustness of the fingerprint database.In the fingerprint matching algorithm,the classic Stack AutoEncoder(SAE)model in deep learning is selected.Compared with traditional machine learning algorithms,deep neural networks have more powerful fitting ability to implement WiFi fingerprint localization algorithm.The pre-trained SAE network is used to reduce the dimension of the processed fingerprint signal,and the extracted high-dimensional fingerprint signals are respectively passed through the classification layer and the regression layer to realize the positioning of the floor and the positioning of the position coordinates within the layer.The effectiveness of the algorithm is verified by experimental simulation.Compared with the traditional algorithm,the proposed algorithm achieves higher positioning accuracy,smaller fingerprint database and lower positioning delay.