Possible future wpa-supplicant behavour question

Robert Moskowitz rgm at htt-consult.com
Fri Jan 21 00:17:58 UTC 2011

I am in the IEEE 802.11ai session.  This is 'Fast Initial Link' setup, a 
new task group in 802.11.

We are looking at how we can accelerate a number of steps in wireless so 
that people walking (or even driving) from one WiFi domain to another 
will actually work.  I am involved in two parts of this work.  To put 
security schemes into the AUTHENTICATE exchanges and IP address 
assignment in the ASSOCIATE exchange.

Let's talk about the IP address assignment.  A scenario would have the 
ASSOCIATE REQUEST have an address request Information Element 
(IE-addr-req); it would have IPv4 and IPv6 flags (so you can get either 
or both back).  An IPv6 IE in the ASSOCIATION RESPONSE would look like 
an IPv6 RA packet.  An IPv4 IE would look like a DHCP response packet.  
This would be pasted up to that kernel in the MLME-MAC-Response, before 
Link establishment.

Here is the question:

Could the kernel accept and handle this information without the Link 
being up?

Given the speed of standards work, we have a couple years to get this 
working  :)

Potentially by the next 802.11 meeting at the end of March, I could have 
diagrams of what the ASSOCIATION frames would look like and it would be 
really great if someone could actually code up a working implementation 
for Proof of Concept.  I am NOT a programmer.

Oh, putting a security exchange in the AUTHENTICATION is on the table as 
well.  I am personally pushing HIP, but we will parametrise it for also 
IKEv2 and 802.1X to establish the WPA MSK, PTK, and GTK (these are the 
various keys used in WPA).

Oh you can see the various presentation on 802.11ai by going to:


and selecting TGai for the currrent efforts and FIA SG for the study 
group efforts


is a presentation I made back in September showing some feasible exchanges.

More information about the test mailing list