|The 2001 Gnutella Research Contest
The Gnutella Network is beginning to attract the attention
of academics and researchers.Lime Wire
LLC would like to encourage further research on the Gnutella Network, as we
believe that such research will benefit the Gnutella community at large.
Therefore, Lime Wire LLC proudly announces
the 2001 Gnutella Research Contest.
Anyone who wishes is invited to submit a research paper to Lime Wire LLC
on a technical issue related to the Gnutella protocol or the Gnutella Network.
One (1) First Place winner will receive a
free trip to New York to discuss his findings with the Lime Wire LLC technical
staff, and will receive $2,000 cash.
Two (2) Second Place winners will receive $500 cash, and ten (10) Third
Place winners will receive Lime Group t-shirts.
Below are some suggestions of technical research topics, but
submissions may be on any topic related to the Gnutella protocol or the
Gnutella Network.For links to
technical resources and papers that have been written about the Gnutella
Network, please click here.
For Official Contest Rules, please click here.
t o p
Currently, the users of the Gnutella Network can set their
number of connections and TTL for their queries.
It would be very useful to know how various combinations of TTL
and number of connections affects message traffic on the network and what
combination of settings is most efficient.
What are the issues with the different configurations?
Schemes for implementing distributed processing on the Gnutella network.
Schemes for indexing the network.
Schemes for improving the reach of queries.
Schemes for network bridging to Gnutella, e.g. FTP and CVS.
Schemes for adding levels of anonymity to the network.
Ideas for dealing with firewalls and NATs.
Ideas for improving download success.
Ideas for new features/services.
Might a "library" model, where materials are "checked out" and made
temporarily unavailable for use by others, be applicable to online file-sharing? How might
such a model be integrated into the Gnutella Network?
What techniques can be used to avoid broadcasting queries? Would
caching of common queries substantially reduce traffic? How about
indexing hosts (like Kazaa), consistent hashing (like Ohaha), or lexical
routing (like Freenet)?
What is a good topology for the Gnutella network? Current discussion
largely ignores the physical organization of the network. But it doesn't
make sense, for example, to have a host in Japan connected to hosts in Texas.
Would transport layers other than TCP (e.g., UDP) significantly help the
scalability of the network? What about IP multicasting?