On the Application of Deep Learning for Precise Indoor Positioning in 6G

TL;DR

This paper explores the use of AI/ML techniques for positioning accuracy enhancement in Indoor Factory (InF) scenarios with a proposed neural network, which is trained on measurements such as Channel Impulse Response and Reference Signal Received Power from multiple Transmit Receive Points (TRPs).

Abstract

Accurate localization in indoor environments is a challenge due to the Non Line of Sight (NLoS) nature of the signaling. In this paper, we explore the use of AI/ML techniques for positioning accuracy enhancement in Indoor Factory (InF) scenarios. The proposed neural network, which we term LocNet, is trained on measurements such as Channel Impulse Response (CIR) and Reference Signal Received Power (RSRP) from multiple Transmit Receive Points (TRPs). Simulation results show that when using measurements from 18 TRPs, LocNet achieves a 9 cm positioning accuracy at the 90th percentile. Additionally, we demonstrate that the same model generalizes effectively even when measurements from some TRPs randomly become unavailable. Lastly, we provide insights on the robustness of the trained model to the errors in ground truth labels used for training.

Figures (6)

• Figure 1: Direct AI/ML Positioning
• Figure 2: AI/ML Assisted Positioning
• Figure 3: Example InF-DH scenario layout with $L = 120m$, $W = 60m$, $D = 20m$, and clutter ($r$, $h_c$, $d_{c}$) = $(60\%, 6m, 2m)$.
• Figure 4: Proposed LocNet model architecture
• Figure 5: Comparison of 90th percentile Horizontal positioning accuracies (test dataset) of LocNet (trained with CIR and CIR+RSRP datasets) versus ResNet variants (ResNet-18, 34, 50, 101) (trained on the CIR+RSRP dataset)