WetLinks: a Large-Scale Longitudinal Starlink Dataset with Contiguous Weather Data

This dataset corresponds to the paper "WetLinks: a Large-Scale Longitudinal Starlink Dataset with Contiguous Weather Data" (https://arxiv.org/abs/2402.16448, accepted for publication in Proc. of the 8th Network Traffic Measurement and Analysis Conf. (TMA))

Please take a look into the paper for detailed info about the measurement setup, software used, as well as analysis results.

Structure

• Raw_Data/ - Contains the unfiltered raw measurement data. The .csv files contain the data and the corresponding .txt files contain descriptions of all features.
  • froggit: Weather data from the Froggit DP2000 weather stations placed directly next to the dishes
  • metadata: detailed version information about the software that we used. This includes, e.g., python modules as well as system libraries
  • net_iperf: the iperf throughput measurement results
  • net_ping: the RTT measurements
  • net_traceroute: the traceroute measurements
  • starlink: Starlink metadata provided by the dish
• preprocessing/ - Scripts that do preprocessing steps on the raw data.
  • throughput_merger.py: Reads the net_iperf.csv file and merges matching measurement entries based on their timestamps. Calculates the mean throughput as well as standard deviation. Then writes the resulting dataframe to a new csv file.
  • csv.merger.py: takes the data from different csv files (net_iperf, net_ping, froggit) and merges them based on their timestamps to one csv can is the input for our analysis.
• Preprocessed_Data/ - Contains the output files from the preprocessing scripts
  • analysis_data: the main file for our analysis. COntains merged throughput, RTT and weather data
  • iperf_cleaned_seconds: contains the second-level iperf measurements. The data is filtered that only measurements running exactly 15 seconds are included. the raw iperf data contains occasional measurements that run significantly longer
  • iperf_cleaned_means: the measurements of iperf_cleaned_seconds are aggregated by their means
• Plotting_Scripts/ - The scripts that create the plots of our paper
• Plots/ - Contains the plots of our paper

DWD and KNMI Weather Data

The weather data from the KNMI (Enschede) and DWD (Osnabrück) are too large to be published on Github. They can be downloaded from the links provided in the "DWD_KNMI_Downloadlinks.txt"

Citation

D. Laniewski, E. Lanfer, B. Meijerink, R. van Rijswijk-Deij, N. Aschenbruck, "WetLinks: a Large-Scale Longitudinal Starlink Dataset with Contiguous Weather Data", Proc. of the 8th Network Traffic Measurement and Analysis Conf. (TMA), 2024.

@inproceedings{datasetpaper,
  title={{WetLinks: a Large-Scale Longitudinal Starlink Dataset with Contiguous Weather Data}},
  author={Laniewski, Dominic and Lanfer, Eric and Meijerink, Bernd and van Rijswijk-Deij, Roland and Aschenbruck, Nils},
  booktitle={Proc. of the 8th Network Traffic Measurement and Analysis Conf. (TMA)},
  year={2024}
}

License

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.