Re: Collecting data to demonstrate TCP ISN-based port knocking

[Christoph Paasch <christoph.paasch@uclouvain.be>]

Hello,

On 14/05/14 - 21:55:36, Julian Kirsch wrote:
> some of you might remember the proposal of a patch which implements a
> variant of port-knocking that can be used to check the authenticity of
> arbitrary TCP connections and even can do integrity checking of TCP
> payload data by using a pre-shared key [0]. This patch, as well as a
> research paper describing its inner workings are available on
> gnunet.org under the name "Knock" [1].
> 
> As Knock uses two fields in the TCP header in order to hide
> information and we explicitly want to be compatible with machines
> sitting in typical home networks, we need to make sure that this
> information doesn't get corrupted by the majority of NAT boxes out
> there. The lack of hard data on this also was one of the objections
> when the patch was submitted last time. We thus created a program
> which tests if Knock could work in your environment. It would be
> greatly appreciated if some of you were able to execute the program on
> their machines in order to help us to get an estimation of if Knock
> one day could be used in a large scale.

have you looked at
http://www.eecs.berkeley.edu/~sylvia/cs268-2014/papers/deploytcp-imc11.pdf ?

Michio started a second larger measurement campaign 2 years ago. You might ask
him if he has more data now.


Cheers,
Christoph