Magnetic-Field-Based Indoor Positioning Using Temporal Convolutional Networks

Abstract

Traditional magnetic-field positioning methods collect magnetic-field information from each spatial point to construct a magnetic-field fingerprint database. During the positioning phase, real-time magnetic-field measurements are matched to a magnetic-field map to predict the user's location. However, this approach requires a significant amount of time to traverse the entire magnetic-field fingerprint database and does not effectively leverage the magnetic-field sequence's unique patterns to improve the accuracy and robustness of the positioning system. In recent years, the application of deep learning for the indoor positioning of magnetic fields has grown rapidly, especially by using the magnetic-field sequence as a time series and a trained long short-term memory (LSTM) model to predict the position, directly avoiding the time-consuming matching process. However, the training of LSTM is time-consuming, and the degradation problem occurs as the stack of layers increases. This article proposes a temporal convolutional network (TCN)-based magnetic-field positioning system that extracts magnetic-field sequence features by preprocessing them with coordinate transformation, smoothing filtering, and first-order differencing. The proposed method is seamlessly applicable to heterogeneous smartphones. The trained TCN models are compared with the LSTM and gated recurrent unit (GRU) models, showing the high accuracy and robustness of the proposed algorithm.

Keywords: heterogenous smartphones; indoor positioning; magnetic field; magnetic trajectories; temporal convolutional networks.

Figure 1

Magnetic field measurements of heterogeneous smartphones in the same corridor. (a) iPhone 12 Mini forward; (b) iPhone 12 Mini backward; (c) iPhone Xs Max forward; (d) iPhone Xs Max backward; (e) Redmi Note 7 forward; (f) Redmi Note 7 backward.

Figure 1

Magnetic field measurements of heterogeneous smartphones in the same corridor. (a) iPhone 12 Mini forward; (b) iPhone 12 Mini backward; (c) iPhone Xs Max forward; (d) iPhone Xs Max backward; (e) Redmi Note 7 forward; (f) Redmi Note 7 backward.

Figure 2

Magnetic field measurement with coordinate transformation in the same corridor. (a) iPhone 12 Mini transformed data forward; (b) iPhone 12 Mini transformed data backward; (c) iPhone Xs Max transformed data forward; (d) iPhone Xs Max transformed data backward; (e) Redmi Note 7 transformed forward; (f) Redmi Note 7 transformed data backward.

Figure 2

Magnetic field measurement with coordinate transformation in the same corridor. (a) iPhone 12 Mini transformed data forward; (b) iPhone 12 Mini transformed data backward; (c) iPhone Xs Max transformed data forward; (d) iPhone Xs Max transformed data backward; (e) Redmi Note 7 transformed forward; (f) Redmi Note 7 transformed data backward.

Figure 3

Visualization of a stack of causal convolutional layers.

Figure 4

Visualization of a stack of dilated causal convolutional layers.

Figure 5

Structure of proposed temporal convolutional networks. (a) Residual learning block; (b) TCN residual block; (c) example of residual connection in a TCN.

Figure 6

Scheme of the indoor magnetic trajectory classification based on a temporal convolutional network.

Figure 7

Designed corridors in Polytech Galilee.

Figure 8

Test trajectory predictions.

Figure 9

Confusion matrix trained and untrained smartphone. (a) Aggregation of three trained smartphones (accuracy: 99.80%); (b) Samsung Galaxy S20 (accuracy: 95.20%); (c) Samsung Galaxy S9 (accuracy: 88.23%); (d) OnePlus 7T Pro (accuracy: 84.27%).