Re: [GNUnet-developers] New NAT Traversal Method

[Bruno Cabral]

Hello,

 

The GNUNet codebase looks weird at first, but can be a good idea to implement your thesis here.

It do have a lot of helpers, and it already have some NAT traversal implementations.

Check :

http://nattest.net.in.tum.de/publications.php

and

https://gnunet.org/pwnat

 

It can be a good baseline in your thesis evaluation.

 

Also, you don’t need to use ICMP hole punching. It is only one technique. Actually it have ICMP, UDP STUN ( new STUN, without TURN) and UPNP.

 

Here you have a good tutorial to get started.

https://gnunet.org/svn/gnunet/doc/gnunet-c-tutorial.pdf

 

[]’s

Bruno

 

 

From: John Michael Lafayette [mailto:address@hidden
Sent: sábado, 22 de agosto de 2015 15:57
To: Bruno Cabral <address@hidden>
Subject: RE: [GNUnet-developers] New NAT Traversal Method

 

Also, I have never implemented or used ICMP hole punching before. Only UDP hole punching. You can use the technique to get around some mobile broadband internet NATs, but it's in its infancy and many of the major providers have chosen TURN over symmetric nat traversal for a number of reasons including wasted packets (many holes have to be punched), potential security risk of open holes being unused, less than 100% success rate, and potentially needing to use modified STUN servers with multiple STUN ports so that the user can tell if their public port number changes when the destination port changes. The last part isn't a necessity, but it is a convenience.

On Aug 22, 2015 2:48 PM, "Bruno Cabral" <address@hidden> wrote:

Hi,

 

The actual STUN code and ICMP NAT Traversal in GNUNet are also written in C.

 

I can help you to implement in GNUnet and test. NAT Traversal is very important for us.

 

[]’s

Bruno

 

 

From: John Michael Lafayette [mailto:address@hidden]
Sent: sábado, 22 de agosto de 2015 15:42
To: Bruno Cabral <address@hidden>
Cc: address@hidden
Subject: Re: [GNUnet-developers] New NAT Traversal Method

 

Dear developer,

I have some WinSocket and Unix socket C code. I couldn't implement it in Java because Java doesn't have the set TTL socket option. I did not test it over mobile broadband yet. I purchased three different mobile broadband aircards (one arrived, two shipping), going to buy some data from T-Mobile, ATT, and have multiple different 3/4g towers in my area to test symmetric NAT traversal on. Need to perfect the technique.

John

 

On 2:34pm, Sat, Aug 22, 2015 Bruno Cabral <address@hidden> wrote:

Hello John,

 

I was responsible for implementing STUN support in GNUNet.

 

Your proposal sounds interesting, did you tried in real NAT ?

 

The actual code can be modified to support it, I can help you to implement it.

 

 

[]’s

Bruno