Icarus is a Python-based simulator for evaluating the performance of in-network caches in Information Centric Networks (ICN).

This document explains how to configure and run the simulator.

To use the simulator, you need:

• Python (v 2.6 onwards, 2.7 preferred): you can either download it from the Python website or, possibly, from the package manager of your operating system
• Python packages: networkx, numpy, fnss. All these three packages are available on PyPi and therefore can be installed using either easy_install or pip.

After you have Python in place with all the required dependencies, you need to download Icarus. You can get it by either cloning the Github repository, running the following command from your command shell:

git clone https://github.com/lorenzosaino/icarus.git

or download the archive file of the repository and unpack it on your machine: [zip] [tar.gz]

Before being able to use it, you need to add the directory where this README file is located to the PYTHONPATH environment variable.

After all these things are taken care of you will be able to use Icarus.

This section explain how to reproduce the results and plot the graphs presented in the paper:

L.Saino, I. Psaras and G. Pavlou, Hash-routing Schemes for Information Centric Networking, in Proc. of the 3rd ACM SIGCOMM workshop on Information Centric Networking (ICN'13), Hong Kong, China, August 2013. [PDF], [BibTex]

To reproduce results and plot the graph, do the following:

1. Open your command shell
2. Move to the directory where this README file is contained
3. Execute the command sh reproduce_hashrouting_icn13_results.sh

This will run all simulations automatically, save all log files and plot all paper graphs.

The logs will be saved in the logs directory, while the graphs will be saved in the graphs directory.

This script has been tested only on Ubuntu 12.04.

If you want to change the configuration of the program, e.g. the range of parameters tested, the number of CPU cores used and so on, please look at the file icarus/config.py. It contains all the configuration parameters and it is well commented.

To use this simulator for gathering results about caching performance wuthout using the script provided, you need to carry out three tasks:

1. Generate the scenarios, i.e. generate the topology and, possibly, event schedule files for all the configurations you want to run
2. Run the actual simulations, once per each set of configuration parameters
3. Plot/analyse the results

To generate the scenarios, you can simply run the file scenario.py located in the directory ./icarus with the parameters already configured. Do:

cd icarus

python scenario.py

If you wish to use different parameters, go and edit the relevant variables in scenario.py. All scenarios are stored in directory ./scenarios.

To run the actual simulations, run the file scenario.py located in the directory ./icarus.

cd icarus

python exec.py

Notice two things: if you open the file you see some configuration variables that you can change. In particular:

• generate_scenarios
• logging.LOG_EVERYTHING

By setting generate_scenarios to True, Icarus automatically generates the scenario files, otherwise, it assumes that the folder ./scenarios already contains the topology files.

logging.LOG_EVERYTHING allows you to enable/disable generation of detailed logs. All logs are stored in the directory ./logs. If this variable is set to False, the simulator will create only two summary files: SUMMARY_CACHE_HIT_RATIO.txt and SUMMARY_NETWORK_LOAD.txt which contain one line per scenario run. If this variable is set to True, then each event of the simulation (cache and link events) will be saved. These file may become very large, in the order of hundreds of GB.

For each single scenario run, if logging.LOG_EVERYTHING configuration variable is set to True, the the simulator generates four log files:

• RESULTS_SCENARIO-ID_CACHE.txt: Tab-separated values, one line per cache event (cache hit, server hit etc..)
• RESULTS_SCENARIO-ID_LINK.txt: Tab-separated values, one line per link event (link X traversed by content Y at time Z etc..)
• RESULTS_SCENARIO-ID_STRETCH.txt (off-path strategies only): Tab-separated values, path stretch for each content retrieval (useful for plotting stretch CDF)
• RESULTS_SCENARIO-ID_DELAY.txt (no cache and CEE+LRU strategies only): Tab-separated values, RTT for each content retrieval (useful for plotting RTT CDF)

In addition, regardless of the value of logging.LOG_EVERYTHING, there simulator generates two summary files:

• SUMMARY_CACHE_HIT_RATIO.txt: each line reports the overall cache hit ratio of a single experiment
• SUMMARY_NETWORK_LOAD.txt: each line reports the overall link load of a single experiment

Each single experiment appends a line to each of these two files.

For further information about the Icarus simulator, please contact Lorenzo Saino