Deep Neural Networks for Indoor Localization Using WiFi Fingerprints

Abstract

In this paper, we propose a novel Wi-Fi positioning method based on Deep Learning. More specifically, we investigate a Stacked AutoEncoder-based model for global location recognition from WiFi fingerprinting data. Stacked AutoEncoder works very well in learning useful high-level features for better representation of input raw data. For our proposed model, two trained unsupervised autoencoders were stacked, then the whole network was trained globally by adding a Softmax output layer for classification. The experimental results show that our Deep Learning based model performs better than SVM and KNN machine learning approaches in a large multi-floor building composed of 162 rooms. Our model achieves an accuracy of 85.58 % and a test time that does not exceed 0.26 s.