i2p.www/i2p2www/pages/site/research/index.html

{% extends "global/layout.html" %}
{% block title %}{{ _('Academic Research') }}{% endblock %}
{% block lastupdated %}{% trans %}May 2018{% endtrans %}{% endblock %}
{% block content %}

<h2>{{ _('Introduction') }}</h2>

<h3>{{ _('Research on the I2P Network') }}</h3>

<p>{% trans -%}
I2P is a very unique project that unfortunately has not received the wider
academic attention it deserves. To date, most research focus on anonymous and
onion-routing technology has been around Tor, and while those papers benefit us
as well, there is a great need for I2P-focused research. This plays a key role
in both maintaining the security and integrity of the network, as well as
opening doors for more impactful future development.
{%- endtrans %}</p>

<p>{% trans anonbib='http://freehaven.net/anonbib/' %}
There is a large research community investigating a wide range of aspects of
anonymity. For a current and comprehensive list of relevant papers, see the
<a href="{{ anonbib }}">Free Haven Anonymity Bibliography</a>.
I2P benefits from much of the research into Tor and onion routing, but there is
little dedicated research interest into the theory behind I2P, and the choices
and tradeoffs that the network makes. This presents a unique opportunity for
original research.
{% endtrans %}</p>

<p>{% trans papers=get_url('papers_list') -%}
A list of known published papers about I2P is available <a href="{{ papers
  }}">here</a>.
{%- endtrans %}</p>

<p>{% trans %}
This page aims to outline the most needed fields of research, notes to
potential researchers, our general safety guidelines as well as an expanding
list of open questions that you can begin on at any time.
{% endtrans %}</p>

<h2>{{ _('Notes to Researchers:') }}</h2>

<h3>{{ _('Defensive Research') }}</h3>

<p>{% trans -%}
While all research on the I2P network is beneficial and appreciated, there are
certain areas which are more in need than others - most so in defensive
research. Most people enjoy coming up with fun ways to launch offensives
against anonymous software, and this is further reinforced by the incentives in
academic institutions. While we know it is often not the first choice for
researchers, We would certainly appreciate any and all work towards ways to
fortify the network!
{%- endtrans %}</p>

<h3>{{ _('Offensive and Analytic Tests') }}</h3>

<p>{% trans contact=site_url('contact') -%}
If you've decided on a research topic that aims to hands-on investigate the I2P
network or solve a problem of large proportions we ask you to <b>please</b> 
<a href="{{ contact }}">communicate your ideas</a> to the development team, the
sooner the better. I2P is under constant development and a significant amount
of roadmapping occurs, therefore your problem may have already been identified
and flagged for update or patch. In the unlikely event you are conducting
testing that overlaps with / would be of interest to another research project
already in motion, we are also able to make you aware of this (with their
permission, of course), and possibly open the door for collaboration. There is
also a chance that the test itself may significantly harm the network or
regular users, and the team may have ideas or suggestions to mitigate that risk
and increase the safety of your testing.
{%- endtrans %}</p>

<h2>{{ _('Research Ethics &amp; Testing the Network') }}</h2>

<h3>{{ _('General Guidelines') }}</h3>

<p>{% trans -%}
<ol>
  <li>
    Consider the benefits and risks - is there any doubt that the research
    provides more value than danger?
  </li>
  <li>
    If the research can be done on a test network then that is the preferred
    method
  </li>
  <li>
    If you must operate on the live network, the safest route is only
    collecting data about yourself
  </li>
  <li>
    If you need 'bigger data', It is recommended to first see if you can use
    data sets from previous experiments or other third party resources is
    recommended
  </li>
  <li>
    If you must collect data on the live network, ensure it is safe for
    publication and collect as little as possible
  </li>
  <li>
    After testing and before publish, review that all data which is to be
    published publicly is <b><i>not</i></b> intended to be private by the
    originator
  </li>
</ol>
{%- endtrans %}</p>

<h3>{{ _('Using a Test Network to Attack I2P') }}</h3>

<p>{% trans -%}
I2P can be run as a separate test network by controlling the locations that a
new router reseeds from so that it only finds other test routers.

The standard mode of operation is to have one JVM per router instance; hence
running multiple copies of I2P on a single machine is inadvisable, both due to
the potential resource drain and the certain port conflicts. To better
facilitate setting up small test networks, I2P has a multirouter mode which
enables multiple distinct routers to be run in the same JVM.

MultiRouter can be started from the i2p base directory by running the below
command.
{%- endtrans %}</p>

<pre>env CLASSPATH=$(find lib/ -name *.jar | paste -s -d ':') java net.i2p.router.MultiRouter 25</pre>

<p>{% trans -%}
Additionally, I2P can be started in a virtual network mode. This mode disables
all transports, allowing the router to be tested in isolation without network
traffic. To enable this mode, add <code>i2p.vmCommSystem=true</code> to the
router.config before starting.
{%- endtrans %}</p>

<h3>{{ _('Testing on the Live I2P Network') }}</h3>

<p>{% trans contact=site_url('contact') -%}
As stated above in the researcher notes, please <a href="{{ contact }}">contact
  us</a> before you commence your testing. While we do not discourage
researchers from responsibly testing their ideas on the live network, if an
attack becomes apparent and we don't have any line of communication then we
will end up taking countermeasures which could interfere with the test.
{%- endtrans %}</p>

<h3>{{ _('Router Family Configuration') }}</h3>

<p>{% trans -%}
As of release 0.9.25, I2P supports a router family configuration. This provides
researchers who run multiple routers with the means to publicly identify those
routers. In turn, this helps the I2P project understand that these routers are
not running an attack on the network. It also will prevent other routers from
including multiple routers of the family in a single tunnel, which could lead
to deanonymization. Routers that appear to be colluding but do not have a
declared family may be assumed to be an attack on the network, and may be
blocked. The best way to ensure the success of your research project is to work
with us directly.
{%- endtrans %}</p>

<p>{% trans -%}
A router family shares a private key so that participation in the family cannot
be spoofed. To configure a router family, click on the 'I2P Internals' link in
the router console, and then on the 'Family' tab. Follow the instructions there
to generate the private key for the first router in the family. Then, export
the key from that router, and import it to other members of the family.
{%- endtrans %}</p>

{% endblock %}