Should counter measures fire for bad MIC during 4-way?

[Jouni Malinen]

On Wed, Jun 06, 2007 at 04:21:57PM -0000, Queisser, Andrew (VfB Stuttgart '07!!) wrote:

> I've been looking at the IEEE spec to answer this question but cannot
> find any verbiage that spells it out so I thought I ask here:
> 
> If a STA sends bad MICs during the 4-way handshake for WPA, should the 
> authenticator go into countermeasures or not?

If you are talking about the initial 4-way handshake and AP detecting
Michael MIC failures, the TKIP keys are only configured after the
handshake and as such, there is not really any chance of detecting a
Michael MIC failure during the handshake.

As far as supplicant reporting Michael MIC failures (i.e., sending
EAPOL-Key with Request and Error flags) is concerned, that could
actually be done immediately after the PTK is derived, i.e., after the
supplicant has received message 1/4. However, in order for this to work,
the attacker would need to be able to derive PTK and that would require
knowing PMK.

I cannot come up with a sequence of events that would allow an attacker
to for countermeasures without having access to PMK or without large
number of calculations to attack RC4 part of TKIP. Did I miss something
here? Can you describe what kind of attack you were thinking about?

-- 
Jouni Malinen                                            PGP id EFC895FA