[VOIPSEC] Actual Attacks

[Johan Bilien]

Le samedi 26 février 2005 à 16:31 +0800, Simon Horne a écrit :

> With reference to the above product, more than ever vendors have to 
> seriously consider Media Encryption with a Handshake technique which foils 
> these types of "wire taps". Methods such as Single Use Diffie Hellman 
> generated half key pairs (with 1 half encrypted) as used in TLS on a 
> seperate secure channel is an excellent method to stop the "Man in the 
> Middle" from being able to decrypt the voice traffic. They may be able to 
> capture to .wav the contents of the conversation but it would be complete 
> garbage. Each conversation or part of conversations are encrypted 
> differently so the 'tapper' has to use repeated blunt force attacks to 
> access the entire conversation. If a large Diffie Hellman "Prime" length is 
> used (> 1536bits) and a high quality cipher (say AES256), makes it almost 
> impossible for all but the the most serious 'tapper' to access.


This is pretty much what the duet MIKEY (RFC3830) / SRTP (RFC3711) tries
to provide. This is implemented in minisip (www.minisip.org), hopefully
other implementations will follow. MIKEY provides a signed D-H exchange
during the call setup (included in the SDP offer/answer), and SRTP uses
the resulting key to encrypt the RTP payload.

Regards,
Johan.