Industrial UWB TWR and TDOA localization dataset in LOS/NLOS scenarios

Citation Author(s):
Van Herbruggen
Ghent University -imec
Vanhie-Van Gerwen
Ghent University - imec
Ghent University - imec
VUB - imec
VUB - imec
University of Antwerp - imec
VUB- imec
Ghent University - imec
De Poorter
Ghent University - imec
Submitted by:
Ben Van Herbruggen
Last updated:
Mon, 04/22/2024 - 16:15
Data Format:
0 ratings - Please login to submit your rating.


This dataset is collected to benchmark different software algorithms for UWB indoor localization. The dataset is collected in a controlled environment with mm-accurate groundtruth to evaluate different algorithms. The data is collected with the Wi-Pos UWB system,a platform developed for research data collection with a wireless long-range Sub-GHz backbone combined with UWB ranging based on the DW1000. The UWB is configured with a center frequency of 6489.6 MHz, a bandwidth of 499.2 MHz (channel 5) and a preamble length of 1024 symbols. The bitrate for the measurements is 850 kbps and a PRF of 64 MHz is used. Eight UWB anchors are placed in a cuboid setup of 8.0 x 10.8 x 2.2 m³ . Both uplink TDoA  and asymmetric double-sided TWR are included in the dataset. For each localization technique, eight different experiments are executed of which eight contain NLOS links (by adding absorbing panels between the tag and anchor nodes). The datasets are collected with a autonomous robot and predefined paths. The predefined paths and groundtruth information of the robot are controlled with a mm-accuracy eight camera MOCAP system. The speed of the autonomous robot varies for the different experiments. For both localization techniques dataset with CIR and without CIR are collected. The update rate of the dataset with CIR is lower due to the amount of data that needs to be collected. From every CIR 300 samples are available. 


This dataset corresponds with the paper: "Selecting and Combining UWB Localization Algorithms: Insights and Recommendations From a Multi-Metric Benchmark" published in IEEE ACCESS.

DOI: 10.1109/ACCESS.2024.3358274


The dataset is structured as follows:

A folder with metadata: anchors coordinates, coordinates of panels, overview of the included experiments.

For each TWR folder: 

- one file with the UWB information: the range, the anchor and time and if avaible (CIR + diagnostics such as first path index, peak path index, fp_ampl, receive power, ....
- one file with the MOCAP data, available at 100 Hz with the timestamp in the same time reference to evaluate the positioning

For each TDOA: 

- one file with the UWB information: timestamp (synchronized to a global UWB time reference) , the anchor and time and if avaible CIR + diagnostics such as first path index, peak path index, fp_ampl, receive power, ....
- one file with the MOCAP data, available at 100 Hz with the timestamp in the same time reference to evaluate the positioning

Funding Agency: 
Grant Number: