I am coding group deaddrops and random packet injection

the deaddrop addresses will be shared by the nodes that are close to it, so just knowing the deaddrop address used (not easy) will not help much. It will also allow all the nodes to be mining these deaddrop addresses!

the one loose end is the node that starts the DHT, it is possible for the traffic pattern to indicate which node started a particular DHT request. At least that is what I would do to try to figure out who is doing what.

The solution is to use the onion routing to randomly pick a node that will inject the DHT request. Since the onion packets will include DHT traffic and non-DHT traffic and they are all the same size, it seems that packet analysis will not be very useful.

Did I miss anything?

James