[Ns-developers] 2009 Google Summer of Code (GSOC) wrapup
tomh at tomh.org
Sun Oct 4 23:08:02 PDT 2009
We had a successful GSOC with all three of our students making nice
contributions to ns-3. I would like to thank all of our students and
mentors, and also Joe Kopena who served as Organizational Admin (OA) and
helped a lot during the selection process. I also would like to
recognize additional mentor volunteers who helped during the selection
process (Nicola Baldo, Marcello Caleffi, and Florian Westphal); we
simply did not have enough spots to take all of the students and mentors
that we would have liked.
Qasim Javed, mentored by Adrian Sai-wah Tam, has contributed Network
Address Translation (NAT) for IPv4. NATs are an important model due to
their pervasive deployment. To implement NAT, however, requires
connection tracking state, and Qasim approached the problem by building
a general framework patterned after Linux Netfilter that can be reused
in the future for things such as firewall rulesets. He then
implemented connection tracking (data structures for maintaining state
about connections) and used the connection tracking API to implement
NAT. At the end of the project, Qasim demonstrated NAT working within
this framework. There remains some integration work before merging to
ns-3-dev, but this code when merged will be a big step forward.
Flavio Kubota, mentored by Juliana Freitag Borin, contributed an uplink
scheduler for WiMAX. WiMAX is a model of great interest to network
simulation, and Flavio was able to complete his primary goal of porting
and validating the uplink scheduler from ns-2 model of the Computer
Networks Laboratory at the Univeristy of Campinas. These extensions are
necessary to model QoS provision on the uplink. Flavio followed through
on the merging of his model to ns-3's WiMAX repository which is being
prepared for our ns-3.7 release.
Duy Nguyen, mentored by Ruben Merz, contributed a new rate control
algorithm for 802.11. Called Minstrel, this rate control is based on a
probing mechanism that seeks to gather statistics about different rates
in the network and apply heuristics to determine the best observed rate.
This approach has been shown to often outperform traditional rate
controls that are based on estimating received signal strength, because
signal strength is only one variable (among channel multipath and
interference) that affects reception probability. Duy's implementation
has already been merged to ns-3-dev, and I'm happy to see his continued
active participation on our users' mailing list.
All of our GSOC work has been archived on the wiki and at our
organizational page here:
In summary, GSOC once again was a great experience for our project and
hopefully for the individuals involved, and we are grateful to Google
for selecting us again this year.
More information about the Ns-developers