[p2p-hackers] Re: Identify "defective" nodes

[Kevin Walsh]

Hi,

There are a lot of proposals for peer-to-peer reputation systems of 
one kind or another, many of them targeted at finding peers that have 
good performance. Unfortunately very few systems have been 
implemented and deployed.

Credence (http://www.cs.cornell.edu/People/egs/credence) is one 
system I have worked on, that has been deployed pretty widely in the 
Gnutella network. The goal of this project is to let clients identify 
spam, decoys, and polluted files, and it can also identify "credible" 
peers in the network. Please feel free to download the source code or 
binary package and play with it, of course.

On the site you will also find some of our papers giving details. The 
papers also have related work and bibliographies listing at least a 
half dozen other proposals that have come up recently in the 
literature. This should let you get a start on peer-to-peer trust, 
reputation, and security literature, in any case.

-Kevin

 > From: ramo at skuff-band.de
 > Subject: [p2p-hackers] Identify "defective" nodes
 >
 > Hi List,
 >
 >
 > i'm a student of Computer Science and trying to search for some good
 > Help's but don't know where to start. So what i want to know are there
 > some techniques for finding defective peers in a , let me say small p2p
 > network with 50 peers. With "defective" i mean , that a peer distribute
 > malicious manipulated services in a service-orientated network, so that
 > the processes on that peers work wrong and is there a technique to
 > eliminate automatically such a peer from the network, i mean something
 > like: put it in a dmz so that it can't do actual bad things.
 >
 > Is there any literature about that or does anyone of the Listpeople know
 > some more about that part?
 >
 > I would be thankful for any Information.
 >
 > Thank you for your thoughts.
 >
 > Ramo