Getting Started

Hello and welcome to the ISAC UAV Dataset from Technische Universität Ilmenau. This page will help you get started with the dataset and provide you with some useful information.

Before downloading the dataset, we want to familiarize you with the structure of the dataset and the provided files.

Please read this section carefully to avoid wasting your costly time in premature self-exploratory efforts.

Dataset organization

The dataset is organized in scenarios. Each scenario contains a series of complex-baseband, time-variant channel transfer functions \(H(f,t)\) recorded at three receivers (Rx). Files that contain sampled transfer-functions use the _channel.h5 suffix.

In addition to the channel transfer functions, the groundtruth position of the UAV is provided in an additional file. Files that contain sampled transfer-functions use the _target.h5 suffix. Hence, each scenario consists of six files:

• 1 file for each Rx (3 in total)
• 3 identical target files (1 for each Rx)

The redundant copies of the target files are provided in case only a single Rx is studied.

For security purposes, the dataset is currently encrypted.

To obtain the password required for the encryption, pleas drop us a short email and state where you are from and how you want to make use of the dataset. A really brief explanation is sufficient :)

Preparations

In its current state, the dataset contains 60 different scenarios in 360 files. The total required disk space is about 200 GB. As Git is not suitable (at least not without extensions) to store such large datasets, the dataset is stored on a different server at TU Ilmenau [1]. The repository provides a script downloader.py to download the dataset from the server.

We tried to keep the requirements minimal, but in case you experience any issues please check the requirements.txt file in the repository.

You can run python -m pip install -r requirements.txt from the repository root to install the required packages.

Downloading the dataset

Clone this repository using the following command:

git clone https://github.com/EMS-TU-Ilmenau/isac-uav-dataset.git

The downloader.py script provides a command-line interface to download the dataset. It takes care of the download, decryption, unpacking, and verification of files. Shasum hashes for the files in the scenarions.shasums also provide a way to track changes of the provided dataset files for the users (via the history of scenarions.shasums).

To download all available files, simply run the downloader.py script without any arguments:

python downloader.py

which will bring you into default mode. In this case, all available scenarios will be downloaded.

The full dataset consumes about 200 GB of disk space. Correspondingly, the time required to download all scenarios will depend on your internet connection. When working with a remote machine, we recommend a screen or tmux session.

Alternatively, you can also specify scenarios with:

python downloader.py --scenario SCENARIO_1 SCENARIO_2 ... SCENARIO_N

Pass the desired scenarios separated by spaces, e.g.,

python downloader.py --scenario 1to2_H15_V11 2to3_H15_V11

To see all available options run:

python downloader.py --help

If you entered a wrong password downloader.py, just re-run the script. In this case, it will not re-download the file and instead access the previously downloaded version stored in the .tmp folder.

Footnote

[1] - We want to thank Henning Schwanbeck from the Technische Universität Ilmenau Rechenzentrum for his continuous support with the provisioning of the measurement files.