May 2003 doc IEEE 802 11 03341 r

May 2003 doc. : IEEE 802. 11 -03/341 r 0 Concerns on EDCF Admission Control Shugong Xu Sharp Laboratories of America, Inc. E-Mail: sxu@sharplabs. com Submission 1 Shugong Xu, Sharp Labs

May 2003 doc. : IEEE 802. 11 -03/341 r 0 Summary • Existing issues in current EDCF admission control proposal. • What options we have? • Straw poll Submission 2 Shugong Xu, Sharp Labs

May 2003 doc. : IEEE 802. 11 -03/341 r 0 Existing issues in the current DAC • Many people in this group think it should be removed and/or replaced. AFAIK, The reason includes: – complexity concern from chip-makers – instability (not quit understand though, since no results shown so far) – difficulty in working with DLP ( the biggest existing hole, in my view, which can be addressed. ) • should be addressed in 11 e • what options we have? Submission 3 Shugong Xu, Sharp Labs

May 2003 doc. : IEEE 802. 11 -03/341 r 0 Option 1: Just remove it without replacement • NO new EDCF admission control will be introduced; • However, many people in this group think EDCF Qo. S should be more than just differentiation between traffics. – Which means user will expect same level of service between now and then, if running an application using EDCF. – Diffserv model from IETF can only work in light load situations since the space for differentiation in 11 is very limited. • Then no way to protect the existing Qo. S traffics – Document 02/544 r 0 demonstrated this. – What the EDCF Qo. S means then? More no-vote may be caused from this option. Submission 4 Shugong Xu, Sharp Labs

May 2003 doc. : IEEE 802. 11 -03/341 r 0 Option 2: replace it without explicit signaling • Using TSPEC, same way as for Polling-based access • Then we will have connection-based EDCF – setup, tear-down, time out, etc – kind of scary? We do not mind. : -) • Some may think it overkill using TSPEC – Why not just use polling based access if AP knows the desires of the STAs? – But why two different signaling systems if defining another signaling, like PSPEC? Sounds like more unacceptable to some folks. Submission 5 Shugong Xu, Sharp Labs

May 2003 doc. : IEEE 802. 11 -03/341 r 0 Option 3: replace it with some unknown-yet mechanism • Leave a hook, which allows the implementation decide – what kind of hook? • how those potential mechanisms work together will be a challenge. Submission 6 Shugong Xu, Sharp Labs

May 2003 doc. : IEEE 802. 11 -03/341 r 0 Straw poll • Option 1: Just remove it ! • Option 2: replace it with explicit signaling as in polling based access • option 3: replace it with some unknown-yet thing • option 4: do nothing • option 5: make it optional Submission 7 Shugong Xu, Sharp Labs