Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CSI-Based User Positioning, Channel Charting, and Device Classification with an NVIDIA 5G Testbed

Reinhard Wiesmayr, Frederik Zumegen, Sueda Taner, Chris Dick, Christoph Studer

TL;DR

This paper introduces CAEZ, three real-world 5G NR CSI datasets collected on a NVIDIA ARC-OTA testbed at ETH Zurich, enabling neural UE positioning, channel charting, and device classification. It presents a unified machine-learning pipeline to process CSI, and demonstrates high-accuracy neural positioning (0.6 cm indoors, 5.7 cm outdoors), channel charting grounded in real-world coordinates (73 cm MAE), and robust device classification across days (up to 99% same-day accuracy). The work fills a critical gap by providing real uplink 5G NR data with ground-truth labels and accompanying code, fostering reproducibility and benchmarking for CSI-based sensing in 5G/6G research. Public availability of the datasets and tools supports rapid development and validation of sensing algorithms in licensed-spectrum 5G NR systems.

Abstract

Channel-state information (CSI)-based sensing will play a key role in future cellular systems. However, no CSI dataset has been published from a real-world 5G NR system that facilitates the development and validation of suitable sensing algorithms. To close this gap, we publish three real-world wideband multi-antenna multi-open RAN radio unit (O-RU) CSI datasets from the 5G NR uplink channel: an indoor lab/office room dataset, an outdoor campus courtyard dataset, and a device classification dataset with six commercial-off-the-shelf (COTS) user equipments (UEs). These datasets have been recorded using a software-defined 5G NR testbed based on NVIDIA Aerial RAN CoLab Over-the-Air (ARC-OTA) with COTS hardware, which we have deployed at ETH Zurich. We demonstrate the utility of these datasets for three CSI-based sensing tasks: neural UE positioning, channel charting in real-world coordinates, and closed-set device classification. For all these tasks, our results show high accuracy: neural UE positioning achieves 0.6cm (indoor) and 5.7cm (outdoor) mean absolute error, channel charting in real-world coordinates achieves 73cm mean absolute error (outdoor), and device classification achieves 99% (same day) and 95% (next day) accuracy. The CSI datasets, ground-truth UE position labels, CSI features, and simulation code are publicly available at https://caez.ethz.ch

CSI-Based User Positioning, Channel Charting, and Device Classification with an NVIDIA 5G Testbed

TL;DR

This paper introduces CAEZ, three real-world 5G NR CSI datasets collected on a NVIDIA ARC-OTA testbed at ETH Zurich, enabling neural UE positioning, channel charting, and device classification. It presents a unified machine-learning pipeline to process CSI, and demonstrates high-accuracy neural positioning (0.6 cm indoors, 5.7 cm outdoors), channel charting grounded in real-world coordinates (73 cm MAE), and robust device classification across days (up to 99% same-day accuracy). The work fills a critical gap by providing real uplink 5G NR data with ground-truth labels and accompanying code, fostering reproducibility and benchmarking for CSI-based sensing in 5G/6G research. Public availability of the datasets and tools supports rapid development and validation of sensing algorithms in licensed-spectrum 5G NR systems.

Abstract

Channel-state information (CSI)-based sensing will play a key role in future cellular systems. However, no CSI dataset has been published from a real-world 5G NR system that facilitates the development and validation of suitable sensing algorithms. To close this gap, we publish three real-world wideband multi-antenna multi-open RAN radio unit (O-RU) CSI datasets from the 5G NR uplink channel: an indoor lab/office room dataset, an outdoor campus courtyard dataset, and a device classification dataset with six commercial-off-the-shelf (COTS) user equipments (UEs). These datasets have been recorded using a software-defined 5G NR testbed based on NVIDIA Aerial RAN CoLab Over-the-Air (ARC-OTA) with COTS hardware, which we have deployed at ETH Zurich. We demonstrate the utility of these datasets for three CSI-based sensing tasks: neural UE positioning, channel charting in real-world coordinates, and closed-set device classification. For all these tasks, our results show high accuracy: neural UE positioning achieves 0.6cm (indoor) and 5.7cm (outdoor) mean absolute error, channel charting in real-world coordinates achieves 73cm mean absolute error (outdoor), and device classification achieves 99% (same day) and 95% (next day) accuracy. The CSI datasets, ground-truth UE position labels, CSI features, and simulation code are publicly available at https://caez.ethz.ch

Paper Structure

This paper contains 29 sections, 10 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Contributions
  3. Relevant Prior Art
  4. ETH Zurich 5G NR Testbed
  5. CAEZ: CSI Acquisition at ETH Zurich
  6. CAEZ-5G-INDOOR
  7. Overview
  8. Measurement Details
  9. CAEZ-5G-OUTDOOR
  10. Overview
  11. Measurement Details
  12. CAEZ-5G-DEV-CLASS
  13. Overview
  14. Measurement Details
  15. Case Studies of CSI-Based Sensing
  16. ...and 14 more sections

Figures (10)

  • Figure 1: Photo of the ETH Zurich 5G NR testbed with and . The 5G network stack runs in real-time on the NVIDIA system nvidia_ARC-OTA_2025. The WorldViz PPT worldviz is used for position tracking.
  • Figure 2: System diagram of the ETH Zurich 5G NR testbed.
  • Figure 3: Measurement setups (top row) and floor plans (bottom row) of the -5G-INDOOR (left) and -5G-OUTDOOR (right) datasets.
  • Figure 4: Measurement setup for the -5G-DEV-CLASS dataset (a), rotation table with mount (b), and the six measured (c).
  • Figure 5: 5G NR testbed and CSI-based machine-learning pipeline.
  • ...and 5 more figures