Analyzing Real-time Video Delivery over Cellular Networks for Remote Piloting Aerial Vehicles

Aygün Baltaci^1,2, Hendrik Cech¹, Nitinder Mohan¹, Fabien Geyer², Vaibhav Bajpai³, Jörg Ott¹, Dominic Schupke²

¹Technical University of Munich, ²Airbus, ³CISPA Helmholtz Center for Information Security

---

This repository contains the scripts and programs that we developed for our publication “Analyzing Real-time Video Delivery over Cellular Networks for Remote Piloting Aerial Vehicles” that was presented at the ACM Internet Measurement Conference 2022. It contains scripts to generate the figures contained in our paper, the scripts that drove our measurements, and the parsers to assemble the reported metrics.

Video delivery pipeline

The application that we used to deliver video with an adaptive bitrate over LTE is available at hendrikcech/imc22-video-pipeline along with setup and usage instructions. It makes use of the gst-timecode GStreamer element.

Dataset

The dataset that we built by conducting drone flights is available for download at mediaTUM. It contains an SQLite database where we aggregated the collected metrics. We ask you to cite our IMC'22 paper upon use.

@inproceedings{uavIMC2022,
    author={Baltaci, Ayg{\"u}n and Cech, Hendrik and Mohan, Nitinder and Geyer, Fabien and Bajpai, Vaibhav and Ott, J{\"o}rg and Schupke, Dominic},
    title={{Analyzing Real-time Video Delivery over Cellular Networks for Remote Piloting Aerial Vehicles}},
    booktitle={Proceedings of Internet Measurement Conference},
    publisher={Association for Computing Machinery},
    address={New York, NY, USA},
    url={https://doi.org/10.1145/3517745.3561465},
    doi={10.1145/3517745.3561465},
    year={2022},
    series={IMC '22}
}

Paper Figures

The figures contained in our paper can be generated with the following steps.

Step 1: Download the dataset

Please visit our mediaTUM publication and download our dataset. It contains a SQLite database from which the plotted data is retrieved. The dataset can be conveniently downloaded using rsync:

export RSYNC_PASSWORD=m1687221
rsync -r –-info=progress2 rsync://[email protected]/m1687221/ imc22-dataset/

Step 2: Set up Python

We used Python version 3.10.5 to generate the plots. Install the required packages listed in requirements.txt either system-wide or in a virtual environment as described below.

pip3 -m venv venv
source venv/bin/activate
pip3 install -r requirements.txt

Step 3: Run the scripts

Each plot is generated by a separate script contained in figures/. To generate all figures, run plot_all.sh with the location of the extracted SQLite database as the first argument. You will find the generated figures in a new directory called figures_pdf.

Measurement scripts

The directory measurements/ contains the scripts that we used to collect data during our UAV flights. Some scripts open a SSH connection to the remote server, using the SSH alias aws. The unique measurement device id is read from /boot/pi_id which is a text file containing a number.

measurements/
├── helpers
│   ├── info.sh
│   ├── iso_date.sh
│   └── tcpdump.sh
├── iperf_server.sh 
├── iperf.sh         # Executed on the client. Uses SSH to execute
│                    # iperf_server.sh on the server.
│
├── ping.sh          # Executes ICMP or TCP pings.
├── roq_play.sh
├── roq_serve.sh     # Executed on the client. Starts roq_play.sh on the server.
├── test-roq.sh      # Starts tcpdump and roq_serve.sh with the selected 
│                    # CC algorithm (static, GCC, or SCReAM).
└── test-utils.sh

Measurement log processing

The folder parsers contains scripts to parse the measurement artifacts that are available at mediaTUM. The tool GNU parallel can be useful to speed up the computation of metrics. By following the listed steps, the relevant metrics are computed and finally compiled into a SQLite database that can be queried easily.

In the following examples, we use two environment variables:

DATASET=./imc22-dataset  # Path to the downloadeded mediaTUM dataset
SCRIPTS=./imc22-scripts  # Path to this repository

Compute SSIM

The SSIM is computed by comparing the source video and the received video. The computation can take a while so we recommend to run multiple computations in parallel. While running, the SSIM results are written next to the receiver video file.

parallel -j 20 --eta "$SCRIPTS/parsers/ssim.py" "$DATASET/video/train_30.mp4" "{}" --method qr --save ::: "$DATASET"/220*/*/*server.avi

Compute latency of RTP and ping packets

ping_pcaps() {
    PCAPS_PI="$(find "$DATASET"/220* -iname '*ping*pi.pcap')"
    for PCAP_PI in $PCAPS_PI; do
        cd "$(dirname "$PCAP_PI")"
        PCAPS_SERVER=( *server*pcap ) # use first pcap
        echo "$PCAP_PI"
        echo "$(pwd)/$PCAPS_SERVER"
    done
}
export -f ping_pcaps

parallel -N 2 "$SCRIPTS/parsers/pcap_latency_loss.py" "{1}" "{2}" ::: \
    $(ping_pcaps)

parallel -N 2 "$SCRIPTS/parsers/pcap_latency_loss.py" "{1}" "{2}" ::: \
    "$DATASET"/220*/*/*roq*rtp.pcap

Populate the SQLite database

Import all csv files into a new SQLite database db.sqlite3.

find "$DATASET"/220* -type f -iname '*csv' \
    -exec "$SCRIPTS/parsers/roq.py" "{}" db.sqlite3 \;

# Populate table 'ping'
find "$DATASET"/220* -type f -iname '*ping*txt' \
    -exec "$SCRIPTS/parsers/roq.py" "{}" db.sqlite3 \;

# Populate table 'flight'
find "$DATASET/flight_logs" -type f -iname '*csv' \
    -exec "$SCRIPTS/parsers/flight.py" "{}" db.sqlite3 \;

# Populate table 'modem'
for PI in 1 2 3 4 5 6; do
    find "$DATASET/lte_ue_logs" -type f -iname "*_${PI}_*log" \
        -exec "$SCRIPTS/parsers/modem_log.py" db.sqlite3 "$PI" "{}" \;
done

# Populate table 'handovers'
for PI in 1 2 3 4 5 6; do
    for MNO in P1 P2; do
        find "$DATASET/lte_ue_logs" -type f -iname "*${PI}_${MNO}.pcap" \
            -exec "$SCRIPTS/parsers/modem_pcap.py" "$PI" "$MNO" "{}" db.sqlite3 \;
    done
done

Add information to the “notes” column of the static tests.

UPDATE `index` SET notes = '5 Mbps'  WHERE day='220217U' AND flight=4 AND pi=6;
UPDATE `index` SET notes = '10 Mbps' WHERE day='220217U' AND flight=5 AND pi=6;
UPDATE `index` SET notes = '15 Mbps' WHERE day='220217U' AND flight=6 AND pi=6;
UPDATE `index` SET notes = '20 Mbps' WHERE day='220217U' AND flight=7 AND pi=6;

UPDATE `index` SET notes = '15 Mbps' WHERE day='220313U' AND flight=1 AND pi=6;
UPDATE `index` SET notes = '25 Mbps' WHERE day='220313U' AND flight=2 AND pi=6;
UPDATE `index` SET notes = '25 Mbps' WHERE day='220313U' AND flight=5 AND pi=6;
UPDATE `index` SET notes = '25 Mbps' WHERE day='220313U' AND flight=9 AND pi=6;

UPDATE `index` SET notes = '10 Mbps' WHERE day='220323R' AND flight=2 AND pi=5;
UPDATE `index` SET notes = '10 Mbps' WHERE day='220323R' AND flight=2 AND pi=6;

UPDATE `index` SET notes = '25 Mbps' WHERE day='220323R' AND flight=3 AND pi=5;
UPDATE `index` SET notes = '8 Mbps' WHERE day='220323R' AND flight=3 AND pi=6;

UPDATE `index` SET notes = '25 Mbps' WHERE day='220323R' AND flight=6 AND pi=5;
UPDATE `index` SET notes = '8 Mbps'  WHERE day='220323R' AND flight=6 AND pi=6;

UPDATE `index` SET notes = '8 Mbps'  WHERE day='220323R' AND flight=12 AND pi=6;
UPDATE `index` SET notes = '12 Mbps' WHERE day='220323R' AND flight=13 AND pi=6;