3

I see from the paper that the complexity of FPC is nlog(n)^2, although it has come up in multiple conversations how FPC can handle a spamming of double spends?

Obviously with enough computational power its possible, but I am curious about what the complexity is, or simply any idea on how FPC can handle a DDOS attack.

If anybody knows anything on the subject it is much appreciated. We could also simply emperically test this on the dev net right now if anybody had the computational resources...

Tsangares
  • 808
  • 5
  • 13

1 Answers1

4

Which FPC paper are you referring to specifically? Let me try to answer this in two steps:

  1. Regarding the double spend spam: in order to communicate regarding the opinion on a conflict transaction one has to only send the identifier of the transaction which is much smaller than the transaction itself. Furthermore one can compress the identifier because some prefix is in most cases sufficient to identify the transaction.

  2. Regarding DDOS In https://arxiv.org/abs/2006.00928 we discuss a method to protect more important nodes against eclipse attacks. These protected nodes issue messages to the Tangle via their gossip channel as special data messages. By doing so they are not required to share their IP and open a port for FPC direct queries, as is suggested in the earlier FPC papers. Effectively disabling this DDOS attack vector.

Andreas Penzkofer
  • 56
  • 1
  • Thank you! My memory was super wrong, for nlog(n)^2 I was referencing Corollary 4.3 in https://arxiv.org/abs/1905.10895 for berserk nodes. You answered this question better than I asked it. – Tsangares Jun 10 '21 at 19:16