Stage-Wise and Hierarchical Training of Linked Deep Neural Networks for Large-Scale Multi-Building and Multi-Floor Indoor Localization Based on Wi-Fi Fingerprinting

This paper present a new solution to the problem of large-scale multi-building and multi-floor indoor localization based on linked deep neural networks (DNNs) - each of which is dedicated to a sub-estimation problem (i.e., building/floor and floor-level location) - trained under the stage-wise and hierarchical training framework. The proposed hierarchical stage-wise training framework extends the original stage-wise training framework to the case of multiple networks by training the DNN for the estimation of floor-level location based on the prior knowledge gained from the training of the DNN for the estimation of building and floor identifiers. The experimental results, with the publicly-available UJIIndoorLoc multi-building and multi-floor Wi-Fi fingerprint database, demonstrate that the linked DNNs trained under the newly-proposed stage-wise and hierarchical training framework can achieve a three-dimensional localization error of 8.19 m, which, to the best of the authors' knowledge, is the most accurate results obtained for the whole of the UJIIndoorLoc database based on DNN-based models.