SRPT-based Congestion Control for Flows with Unknown Sizes

This repository contains the code necessary to reproduce the results presented in the "SRPT-based Congestion Control for Flows with Unknown Sizes" paper; specifically, Figures 4 and 5.

Depencencies

Recent Clang C++ compiler
The CMake build tool
Tcl/Tk
OTcl
TclCL
spdlog
fmt
Python
Click python library
TOML python library

Building

NS2-simulator

cd ns2
mkdir cmake-build-release
cd cmake-build-release
cmake .. -DCMAKE_BUILD_TYPE=Release
make
cd ../..

Auxiliary oracle-fct executable

cd scripts/plot_generation
mkdir cmake-build-release
cd cmake-build-release
cmake .. -DCMAKE_BUILD_TYPE=Release
make
cd ../../../

Running simulations

Preparation:

export DEV_NS_DIR=$PWD/ns2
cd scripts # change to scripts dir

Running actual simulations:

python -m congestion_runner.run --scale final --control ascc\
    --buffer ascc --input web_search --slacks default --load 0.8 --alpha 2.0
# see the description of CLI arguments below

The simulation results would be available in the results folder.

Simulation CLI description

Below we provide a detailed description of the CLI arguments.

Algorithms:
- PIAS: --buffer pias --control pias
- aSCC: --buffer ascc --control ascc
- pFabric: --buffer pfabric --control pfabric_clairvoyant
- LAS: --buffer las --control pias
Inputs:
- Web Search: --input web_search
- Data Mining: --input data_mining
Parameter specification:
- Load: --load 0.8
- Alpha: --alpha 1.0
Other mandatory arguments:
- --scale final --slacks default

Statistic extraction

Preparation

mkdir plots
export PLOT_PATH=$PWD/plots
cd scripts

Generating plot data in .csv format:

# use same --control --buffer --input and --alpha parameters as for 
# running simulations
python -m plot_generation.generate_plots --aspect load --value 0.8\
    --scale final --control ascc --buffer ascc --input web_search\
    --slacks default --alpha 2.0

The above command would create a .csv fail in the PLOT_PATH dir.