Skip to content

Latest commit

 

History

History

plots

README.md

Scripts to process and plot experiment results

Overview

The scripts in this directory serve the processing and plotting of the experiment results

  • parse_results.py transform the logs from the experiment runs into easier to work with CSV files.
  • plot_cdf.py is used to generate the CDF plots you can see in the paper from the times CSV files.
  • plot_stats.py is used to generate the bar plots you can see in the paper (and many more) from the stats CSV files.
  • plot_scatter.py is used to generate the scatter plots you can see in the paper from the stats CSV files.

Requirements

The scripts assume they are run with Python 3.

The following python packages are required (version numbers indicate tested versions):

  • matplotlib v3.3.0
  • pandas v1.0.0

The required packages are listed in requirements.txt and can be installed using

pip3 install -r requirements.txt

Usage

This scripts takes the logs from your experiments runs and transforms them into easy to digest CSV files. Two files for each combination of topology, scenario, and fragmentation forwarding variant are generated.

  • -times.csv which contains a line for each interest transmitted by the consumer during the experiments, logging its sent time according to provided xtimer timestamp (see app README) and the reception time of the content chunk if received by the consumer. It also contains columns to describe the experiment setup.
  • -stats.csv which contains a line for each node participating in the experiments per run logging all scalar stats for that run as well as additional columns describing the experiment setup.

The script takes optionally an unlimited number of arguments. The arguments are names of logs (without preceding path) that should be ignored:

./parse_results.py [blacklisted logs]

Environment variables

  • DATA_PATH: (default: ./../../results) Path where the logs to consider are stored.

plot_cdf.py

This script plots the CDF of the time-to-completion for each interest transmitted by the consumer. It requires the -times.csv files to take the data from as arguments:

./plot_cdf.py ../../results/*-times.csv

plot_stats.py

This script creates bar plots for various scalar stats for each run and participating node. It requires the -stats.csv files to take the data from as arguments:

./plot_stats.py ../../results/*-stats.csv

Optionally, the kind of plotted stats can be restricted using the -s argument:

./plot_stats.py -s int_retrans,pktbuf ../../results/*-stats.csv
usage: plot_stats.py [-h] [-s [STATS_TO_PLOT]] filenames [filenames ...]

positional arguments:
  filenames             CSV files as generated by ./parse_results.py to takes
                        stats from

optional arguments:
  -h, --help            show this help message and exit
  -s [STATS_TO_PLOT], --stat [STATS_TO_PLOT]
                        Comma separated list of stat to plot. Possible values:
                        cnt_trans, cs_hits, fbuf_full, frag_fwd, frag_retrans,
                        int_retrans, pktbuf, vrb_full, rbuf_full. Default: all

plot_scatter.py

This script creates scatter plots for the various scalar stats for each run also handled in ./plot_stats.py for a specified node. It requires the a pair of the desired stats, the node and -stats.csv files to take the data from as arguments:

./plot_scatter.py ./plot_scatter.py m3-281 cs_hits vrb_full \
    ../../results/*-stats.csv
usage: plot_scatter.py [-h] node stat1 stat2 filenames [filenames ...]

positional arguments:
  node        Node to plot scatter plot for
  stat1       Stat for x-axis
  stat2       Stat for y-axis
  filenames   Filenames

optional arguments:
  -h, --help  show this help message and exit