A Novel Indoor Positioning Based on Geomagnetic Trajectory Signal

Ubiquitous geomagnetic field exhibits unique features due to differences between building structures in indoor environments. In addition, the difficulty of resolving geomagnetic signals will lead to inaccurate positioning results. This paper proposes a geomagnetic indoor positioning system using deep neural networks to improve positioning accuracy. In order to solve the problem of low discernibility of the geomagnetic filed, continuous geomagnetic signals are vectorized into a trajectory sequence, and a novel map-building method is designed based on this to geomagnetic database for indoor positioning. Then, temporal convolutional network (TCN) is introduced to extract deep features of geomagnetic trajectory sequences. Experimental results show that the method outperforms other machine learning algorithms, such as KNN and LSTM-based DRNN.