Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TDoA-Based Self-Supervised Channel Charting with NLoS Mitigation

Mohsen Ahadi, Omid Esrafilian, Florian Kaltenberger, Adeel Malik

TL;DR

The paper presents a self-supervised channel charting framework that leverages CIR data, TDoA measurements, and known TRP locations to achieve global 2D UE embeddings without ground-truth labels. By incorporating displacement-derived continuity and an NLoS masking mechanism, the approach improves robustness in mixed LoS/NLoS environments. The method is evaluated in Matlab simulations and on a real GEO-5G testbed with O-RAN/OpenAirInterface, achieving 2–4 m localization accuracy at 90% of cases across varying LoS ratios and outperforming state-of-the-art CC baselines. Public CIR datasets and ground-truth positions are released to support further research, and future work includes deeper E2 integration and real-time RIC deployment.

Abstract

Channel Charting (CC) has emerged as a promising framework for data-driven radio localization, yet existing approaches often struggle to scale globally and to handle the distortions introduced by non-line-of-sight (NLoS) conditions. In this work, we propose a novel CC method that leverages Channel Impulse Response (CIR) data enriched with practical features such as Time Difference of Arrival (TDoA) and Transmission Reception Point (TRP) locations, enabling a self-supervised localization function on a global scale. The proposed framework is further enhanced with short-interval User Equipment (UE) displacement measurements, which improve the continuity and robustness of the learned positioning function. Our algorithm incorporates a mechanism to identify and mask NLoS-induced noisy measurements, leading to significant performance gains. We present the evaluations of our proposed models in a real 5G testbed and benchmarked against centimeter-accurate Real-Time Kinematic (RTK) positioning, in an O-RAN--based 5G network by OpenAirInterface (OAI) software at EURECOM. It demonstrated outperforming results against the state-of-the-art semi-supervised and self-supervised CC approaches in a real-world scenario. The results show localization accuracies of 2-4 meters in 90% of cases, across a range of NLoS ratios. Furthermore, we provide public datasets of CIR recordings, along with the true position labels used in this paper's evaluation.

TDoA-Based Self-Supervised Channel Charting with NLoS Mitigation

TL;DR

The paper presents a self-supervised channel charting framework that leverages CIR data, TDoA measurements, and known TRP locations to achieve global 2D UE embeddings without ground-truth labels. By incorporating displacement-derived continuity and an NLoS masking mechanism, the approach improves robustness in mixed LoS/NLoS environments. The method is evaluated in Matlab simulations and on a real GEO-5G testbed with O-RAN/OpenAirInterface, achieving 2–4 m localization accuracy at 90% of cases across varying LoS ratios and outperforming state-of-the-art CC baselines. Public CIR datasets and ground-truth positions are released to support further research, and future work includes deeper E2 integration and real-time RIC deployment.

Abstract

Channel Charting (CC) has emerged as a promising framework for data-driven radio localization, yet existing approaches often struggle to scale globally and to handle the distortions introduced by non-line-of-sight (NLoS) conditions. In this work, we propose a novel CC method that leverages Channel Impulse Response (CIR) data enriched with practical features such as Time Difference of Arrival (TDoA) and Transmission Reception Point (TRP) locations, enabling a self-supervised localization function on a global scale. The proposed framework is further enhanced with short-interval User Equipment (UE) displacement measurements, which improve the continuity and robustness of the learned positioning function. Our algorithm incorporates a mechanism to identify and mask NLoS-induced noisy measurements, leading to significant performance gains. We present the evaluations of our proposed models in a real 5G testbed and benchmarked against centimeter-accurate Real-Time Kinematic (RTK) positioning, in an O-RAN--based 5G network by OpenAirInterface (OAI) software at EURECOM. It demonstrated outperforming results against the state-of-the-art semi-supervised and self-supervised CC approaches in a real-world scenario. The results show localization accuracies of 2-4 meters in 90% of cases, across a range of NLoS ratios. Furthermore, we provide public datasets of CIR recordings, along with the true position labels used in this paper's evaluation.

Paper Structure

This paper contains 21 sections, 21 equations, 23 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. O-RAN Network Architecture
  4. Data Pre-Processing and Feature Extraction
  5. TDoA Channel Charting with Sensor Fusion and NLoS Mitigation
  6. TDoA Channel Charting with Sensor Fusion and NLoS Mitigation
  7. Training Channel Charting Models
  8. TDoA Loss Function
  9. TDoA+Displacement Loss Function
  10. NLoS Classification Function
  11. Testing Channel Charting Models
  12. Evaluations and Results
  13. Matlab Simulations
  14. GEO-5G Testbed
  15. Synchronization
  16. ...and 6 more sections

Figures (23)

  • Figure 1: 5G AI/ML Positioning system model.
  • Figure 2: CIR shifting, truncation, and normalization in pre-processing while keeping the TDoA integrity
  • Figure 3: CIR pre-processing diagram
  • Figure 4: CC training with CIR and TDoA
  • Figure 5: CC training with CIR and TDoA+displacement
  • ...and 18 more figures