On Sat, Nov 09, 2002 at 09:15:39PM -0500, Scott Young wrote: > HTL has 2 main problems: > First hop - Incoming HTL values can be used to determine if a node is likely > to be the originator of a request > HTL-1 - Malicious node may find out if a node contains specific data.
The reason such changes have never had very high priority here is that the HTL does not reveal very much which cannot be seen in other ways. For example, if you keep track of what keys your node has released references for, you can tell pretty well if you are at the first hop by the proximity of the search key to the references you have given (especially since the first hop is currently randomly routed). If it doesn't match a reference in the first 10 bits or so, then it is pretty sure you have a first hop. At the other end, timing will give away whether the data was on the node you queried or whether it forwarded the request. Generally, both problems are more or less intractable. The very nature of a routing algorithm is that you stepwise inprove the accuracy of the search - low accuracy always indicates an early phase. And while you can add random delays at the last node, a statistical attack will always be able to reveal these (though, if one goes with the argument "the second time you request the data it is obviously in the node - you put it there" then that gets more difficult for particular cases.) <> -- Oskar Sandberg oskar at freenetproject.org _______________________________________________ devl mailing list devl at freenetproject.org http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl
