Rate Control Protocol (RCP)

This project was done for CS 244 Spring 2014 by Lisa Yan and Richard Hsu. The project includes simulation code used to reproduce the RCP work results of the Average Completion Time for varying Flow Sizes (See Figure 12 of the full paper).

Rate Control Protocol

"Why Flow-Completion Time is the Right Metric for Congestion Control" [pdf] [full version] by Nandita Dukkipati and Nick McKeown (http://yuba.stanford.edu/rcp/)

Reproduction Setup

Your Own Setup

The following are instructions to help you run it on your own system. We cannot guarantee it will work for all systems but this has been tested on Ubuntu 14.04 LTS.

Installing NS-2.35 with RCP code

You will need to install NS-2.35 All In One package on your system, which you can place anywhere in your system.

Step 0: You'll want to make sure you have the necessary tools installed. The following can be used for Ubuntu systems and have been tested for Ubuntu 14.04 LTS. Generally speaking you will need tools for building, autoconf, g++/gcc and X11 library code.

sudo apt-get install git build-essential autoconf automake perl g++ libx11-dev libxt-dev libx11-dev libxmu-dev xorg-dev xgraph

Step 1: In the folder that you want NS-2.35 code perform the following steps:

wget -O ns-allinone-2.35.tar.gz http://downloads.sourceforge.net/project/nsnam/allinone/ns-allinone-2.35/ns-allinone-2.35.tar.gz?r=http%3A%2F%2Fsourceforge.net%2Fprojects%2Fnsnam%2Ffiles%2Fallinone%2Fns-allinone-2.35%2F&ts=1401492260&use_mirror=tcpdiag
tar -xzvf ns-allinone-2.35.tar.gz

Step 2:: Next you'll want to patch it for RCP. Copy the rcp-2.35.patch file from our repository into the ns-allinone-2.35 directory. Then perform the patch:

patch -p1 < rcp-2.35.patch

Step 3: Now you'll want to install the whole NS package and dependencies. In the top level of the ns-allinone-2.35 folder run the install command:

./install

Hopefully it will install successfully; if not read the error codes and fix. Unfortunately we may not be able to provide assistance with this.

Step 4: The installation should provide some output on changing the path. You should follow those instructions but here it is reproduced assuming the path to ns-allinone-2.35 is /home/ubuntu/ns-allinone-2.35:

Please put /home/ubuntu/ns-allinone-2.35/bin:/home/ubuntu/ns-allinone-2.35/tcl8.5.10/unix:/home/ubuntu/ns-allinone-2.35/tk8.5.10/unix
into your PATH environment; so that you'll be able to run itm/tclsh/wish/xgraph.

IMPORTANT NOTICES:

(1) You MUST put /home/ubuntu/ns-allinone-2.35/otcl-1.14, /home/ubuntu/ns-allinone-2.35/lib,
    into your LD_LIBRARY_PATH environment variable.
    If it complains about X libraries, add path to your X libraries
    into LD_LIBRARY_PATH.
    If you are using csh, you can set it like:
                setenv LD_LIBRARY_PATH <paths>
    If you are using sh, you can set it like:
                export LD_LIBRARY_PATH=<paths>

(2) You MUST put /home/ubuntu/ns-allinone-2.35/tcl8.5.10/library into your TCL_LIBRARY environmental
    variable. Otherwise ns/nam will complain during startup.

An example for what may be put in .bashrc file:

export PATH=$PATH:/home/ubuntu/ns-allinone-2.35/bin:/home/ubuntu/ns-allinone-2.35/tcl8.5.10/unix:/home/ubuntu/ns-allinone-2.35/tk8.5.10/unix

export LD_LIBRARY_PATH=/home/ubuntu/ns-allinone-2.35/otcl-1.14:/home/ubuntu/ns-allinone-2.35/lib

export TCL_LIBRARY=/home/ubuntu/ns-allinone-2.35/tcl8.5.10/library

Python libraries

Our scripts require additional Python libraries to run additional simulations and to generate the graphics. You can install them on Ubuntu as follows:

sudo apt-get install python-numpy python-scipy python-matplotlib ipython ipython-notebook python-pandas python-sympy python-nose

For other systems you can refer to the Matplotlib Documentation or NumPy/SciPy Documentation for more information.

Reproduction Code

Now you just need to go into the simulation folder of our code and type ./run.sh. You may need execution permissions if it isn't set properly when you've cloned the repository. You can find the graphs produced in the simulation/graphs , which reproduce Figure 12 in the full version of the RCP paper. Note that the simulations and plotting will take roughly 1.5 hours to complete and you should use screen or tmux so that any issues of disconnecting from the server do not interrupt the run.

richardhsu/cs244-14-rcp

Rate Control Protocol (RCP)