Capwap Mailing List: RE: question on LWAPP-03 draft

[Pat Calhoun (pacalhou) (pacalhou]

Hopefully the following answers the questions to your satisfaction.
Please let me know if you have any additional comments/questions.

(a) That is precisely the problem that I have been raising regarding
encryption in the AC. Fragmentation rarely (if ever) occurs today. Sure,
it's in the 802.11 spec but no one does it. However, with the
introduction of HCCA fragmentation will become commonplace in order to
satisfy a service period. Therefore, fragmentation will have to occur on
the WTP (which has real-time access to RF conditions), and as a
consequence encryption will have to happen on the WTP.

(b)Good question. The Sequence Number subfield is managed by the WTP,
not the AC. I will need to add that clarification to the next rev of the
document. 

Thanks,

Pat Calhoun
CTO, Wireless Networking Business Unit
Cisco Systems

________________________________

From: Puneet Agarwal [mailto:pagarwal [at] broadcom.com] 
Sent: Saturday, July 23, 2005 3:44 PM
To: capwap [at] frascone.com; Pat Calhoun (pacalhou)
Subject: question on LWAPP-03 draft



        Hi Pat,
         
        I had a couple of question on the "Division of Labor" in Split
MAC (section 11.1.1) in draft-ohara-capwap-lwapp-03.txt:
         
        (a) In the draft, it is stated that the 802.11i
"Encryption/Decryption" can happen either on the WTP or on the AC. It
further states that the Fragmentation/Defragmentation occurs on the WTP.
If one considers the case when Encryption/Decryption happens on the AC,
then for a 802.11 fragment, I do not see how the defragmentation can
happen on the WTP. Each 802.11 fragment (of an MSDU) needs to be
decrypted before it can be defragmented. The one possible way I can see
this happening would be Client--> WTP--->AC(decrypt)---->
WTP(defrag)--->AC--->network but I presume that this is not what you
intended. 

         
        (b) Secondly, who maintains the counter for the "Sequence
Number" sub-field in the "Sequence-Control" field for frames transmitted
to the Client? Is it the AC or the WTP? It would seem that for split
MAC, this should really be maintained at the AC. 
        However in your Split MAC model you mention that "Beacon
generation" is a WTP functionality which would imply that this counter
must really be maintained in the WTP. This would imply that the
"Sequence Number" sub-field in the 802.11 frames tunneled from the AC to
WTP is effectively ignored by the WTP. Is this the intended behavior?