1. MAC improvement using random AIFSN
May 2006
doc.: IEEE /0657r3
July 2006
MAC improvement using random AIFSN
Date:
Authors:
Notice: This document has been prepared to assist IEEE It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein.
Release: The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this contribution may be made public by IEEE
Patent Policy and Procedures: The contributor is familiar with the IEEE 802 Patent Policy and Procedures < ieee802.org/guides/bylaws/sb-bylaws.pdf>, including the statement "IEEE standards may include the known use of patent(s), including patent applications, provided the IEEE receives assurance from the patent holder or applicant with respect to patents essential for compliance with both mandatory and optional portions of the standard." Early disclosure to the Working Group of patent information that might be relevant to the standard is essential to reduce the possibility for delays in the development process and increase the likelihood that the draft publication will be approved for publication. Please notify the Chair as early as possible, in written or electronic form, if patented technology (or technology under patent application) might be incorporated into a draft standard being developed within the IEEE Working Group. If you have questions, contact the IEEE Patent Committee Administrator at
Todor Cooklev
Todor Cooklev

2. Outline Contention in 802.11 scenarios for digital home
May 2006
doc.: IEEE /0657r3
July 2006
Outline
Contention in scenarios for digital home
Current e channel access mechanism
Importance of AIFSN
Why is random AIFSN better than fixed?
Proposed channel access mechanism
Simulation results
Implementation considerations
Conclusions
Todor Cooklev
Todor Cooklev

3. Contention in 802.11 scenarios for digital home
May 2006
doc.: IEEE /0657r3
July 2006
Contention in scenarios for digital home
Many traffic streams anticipated in CE dominated home…
HD Video
Support for 4-5 HD video streams Mbps
Content: Premium, time-shifted, DVD, personal, Gaming
Voice
Up to 4 streams with tight delay constraints – low throughput
Best-effort
Web, file transfer, printing - 10Mbps
Contention among equal-priority traffic is an important issue
Todor Cooklev
Todor Cooklev

4. Digital home of the future
May 2006
doc.: IEEE /0657r3
July 2006
Digital home of the future
HD content
HDTV
HDTV V4
HDTV V3 V2 V1
Live or time-shifted
Premium HD content
Voice 3-4
Data 2-3
Cable, Satellite
STB / PVR
Media PC
over WLAN
Internet, IPTV
Personal
content V5
Todor Cooklev
Todor Cooklev

5. 802.11e channel access mechanism
May 2006
doc.: IEEE /0657r3
July 2006
802.11e channel access mechanism
Fig.1 EDCA channel access mechanism
Arbitration inter-frame space (AIFS) is one of the EDCA parameters used to ensure traffic differentiation
Each AC has fixed AIFS duration associated with it. The smaller the AIFS the higher the medium access priority
But multiple streams of the same AC use same AIFSN leading to increased collision!
Fig. 2 From Table 20.2 of 11e
Todor Cooklev
Todor Cooklev

6. AIFSN is important for wireless QoS!
May 2006
doc.: IEEE /0657r3
July 2006
AIFSN is important for wireless QoS!
AIFS is
A critical element for providing high priority services under heavily loaded scenarios
Depletes channel access to low priority stations in heavily loaded scenarios
Fair bandwidth sharing
Todor Cooklev
Todor Cooklev

7. AIFS’ Role in Traffic Prioritization
May 2006
doc.: IEEE /0657r3
July 2006
AIFS’ Role in Traffic Prioritization
Scenario
1 AP with 4 STAs
AIFS of STA 1,2 3,4 set to [2,3,4,5] respectively.
[CWmin CWmax] = [15,31] for all STAs.
AIFSN = 2
AIFSN = 3
Small differences in AIFS translates to large differentiation between flows
Fixed AIFSN value does not guarantee finer level of prioritizations.
Fixed AIFSN value for multiple STAs can lead to increased collisions
AIFSN = 4
AIFSN = 5
Todor Cooklev
Todor Cooklev

8. Therefore fixed AIFSN for all stations at the same AC does not help!
May 2006
doc.: IEEE /0657r3
July 2006
How effective is adjusting the AIFSN? (assuming that AIFSN is fixed per AC)
Compare the following two scenarios with CWmin=7 and CWmax=15, AC2
Scenario I: For each QSTA AIFSN=2.
Scenario II: For each QSTA AIFSN=3.
The probability for collisions is identical in both cases!
Therefore fixed AIFSN for all stations at the same AC does not help!
Todor Cooklev
Todor Cooklev

9. Proposed channel access mechanism
May 2006
doc.: IEEE /0657r3
July 2006
Proposed channel access mechanism
AIFSN is random with uniform probability density function in [2 3].
On average, for 5 QSTA AIFSN=2 and for another 5 QSTA AIFSN=3.
Advantages
THE COLLISION PROBABILITY IS REDUCED!
Possible to assign an AC a non-integer AIFSN value on an average
Todor Cooklev
Todor Cooklev

10. Collision probability vs. number of stations
May 2006
doc.: IEEE /0657r3
July 2006
Collision probability vs. number of stations
(random AIFSN=[2 4] and CW=8; fixed AIFSN=4)
Todor Cooklev
Todor Cooklev

11. Random AIFSN=[2, 4] vs fixed AIFSN=4
May 2006
doc.: IEEE /0657r3
July 2006
Relative improvement in collision probability vs number of stations for different CW size
Random AIFSN=[2, 4] vs fixed AIFSN=4
Todor Cooklev
Todor Cooklev

12. Probability vs. number of attempts to send one packet
May 2006
doc.: IEEE /0657r3
July 2006
Probability vs. number of attempts to send one packet
Random AIFSN=[2, 4]; fixed AIFSN=4
Todor Cooklev
Todor Cooklev

13. Random AIFSN=[2, 4]; fixed AIFSN=4
May 2006
doc.: IEEE /0657r3
July 2006
Ratio between probabilities of success at the first attempt for random and fixed AIFSN
Random AIFSN=[2, 4]; fixed AIFSN=4
Todor Cooklev
Todor Cooklev

14. Simulation Results with TGnSync OPNET Model
May 2006
doc.: IEEE /0657r3
July 2006
Simulation Results with TGnSync OPNET Model
2x2 MIMO, 20MHz Channel, Offered load = 15Mbps/node
Average Retransmissions by N4
Fixed AIFSN = 2
Fixed AIFSN = 3
Random AIFSN [2,3]
CWmin = 7, CWmax = 15
Todor Cooklev
Todor Cooklev

15. Simulation Results with TGnSync OPNET Model
May 2006
doc.: IEEE /0657r3
July 2006
Simulation Results with TGnSync OPNET Model
2x2 MIMO, 20MHz Channel, Offered load = 15Mbps/node
Average Retransmissions by N4
CWmin = 7, CWmax = 15
Fixed AIFSN = 2
Fixed AIFSN = 3
Random AIFSN [2,3]
CWmin=7, CWmax = 31, Fixed AIFSN = 2
Todor Cooklev
Todor Cooklev

16. Simulation Results with TGnSync OPNET Model
May 2006
doc.: IEEE /0657r3
July 2006
Simulation Results with TGnSync OPNET Model
Fixed AIFSN = 2, CWmin=7, CWmax = 31
Average Delay at N5
Fixed AIFSN = 3
Fixed AIFSN = 2
Random AIFSN [2,3]
Average Throughput at N5
Fixed AIFSN = 2
Fixed AIFSN = 3
Todor Cooklev
Todor Cooklev

17. How are other ACs affected?
May 2006
doc.: IEEE /0657r3
July 2006
How are other ACs affected?
Two scenarios for AC2 – fixed AIFSN vs. random AIFSN
10 QSTA, all in AC2, each with fixed AIFSN
5 QSTA, with AIFSN=2 and 5 QSTA with AIFSN=3 (on average), all 10 QSTA in AC2
Which scenario is better for QSTA belonging to other ACs? - Scenario 2 because:
Reduced collisions leads to better bandwidth utilization, which helps all QSTA!
Todor Cooklev
Todor Cooklev

18. Simulation results with random AIFSN
May 2006
doc.: IEEE /0657r3
Simulation results with random AIFSN
July 2006
Simulation Parameters
2x2 MIMO, 20MHz Channel, 5 pairs each of BE, Video and Voice traffic
AIFSN[AC1] – fixed default values as in Table 20.2, except for video; AIFSN for video is random with uniform distribution over [1 3].
Throughput per stream
Delay per stream
Best Effort
Best Effort
Video
Video
Voice
Voice
Todor Cooklev
Todor Cooklev

19. Simulation results with random AIFSN
May 2006
doc.: IEEE /0657r3
Simulation results with random AIFSN
July 2006
Simulation Parameters
54 Mbps PHY (802.11g), 4 pairs each for BE, Video and Voice traffic
AIFSN[AC1] – fixed default values as in Table 20.2, except for video; AIFSN for video is random with uniform distribution over [2 4].
Global Throughput per AC
Global Delay per AC
Best Effort
Best Effort
Video
Video
Voice
Voice
Todor Cooklev
Todor Cooklev

20. Proposed Text : Section 9.2.3.4
May 2006
doc.: IEEE /0657r3
July 2006
Proposed Text : Section
Original…
A Non-AP QSTA computes the time periods for each AIFS[AC] from the dot11EDCATableAIFSN attributes in the MIB. QSTAs update their dot11EDCATableAIFSN values using information in the most recent EDCA Parameter Set element of Beacons received from the QAP of the QBSS (see ). A QAP computes the time periods for each AIFS[AC] from the dot11QAPEDCATableAIFSN attributes in its MIB.
NEW…
Non-AP QSTA compute AIFSN[AC] using uniform probability density function over a finite interval. This interval is bounded on one side by the default values in Table The other limit of this interval is from the dot11EDCATableAIFSN attributes in the MIB. QSTAs update their dot11EDCATableAIFSN values using information in the most recent EDCA Parameter Set element of Beacons received from the QAP of the QBSS (see ).
Todor Cooklev
Todor Cooklev

21. Proposed Text : Section 9.9.1.3
May 2006
doc.: IEEE /0657r3
July 2006
Proposed Text : Section
Original …
The value of AIFSN[AC] shall be greater than or equal to 2 for non-AP QSTAs and is advertised by the QAP in the EDCA Parameter Set Information Element in Beacons and Probe Response frames transmitted by the QAP. The value of AIFSN[AC] shall be greater than or equal to 1 for QAPs.
NEW…
QSTAs select the value of AIFSN according to a uniform probability density function (PDF) over a finite interval. One of these limits (either the lower or the upper limit) is the default value specified in Table The other limit is in the dot11EDCATableAIFSN attribute in the MIB. Any PDF is acceptable, including a PDF where QSTA always select one number that belongs to this interval. The limit value AIFSN[AC] is advertised by the QAP in the EDCA Parameter Set Information Element in Beacons and Probe Response frames transmitted by the QAP. If this limit value is smaller than the default values in Table 20.2, then it becomes the lower bound of the interval. If this limit value is higher than the default values in Table 20.2, then it becomes the higher bound of the interval.
Todor Cooklev
Todor Cooklev

22. Implementation Issues
May 2006
doc.: IEEE /0657r3
July 2006
Implementation Issues
The recommendation is optional.
Since uniform probability density function is recommended, all current implementations using fixed AIFSN is a special case of this proposal, i.e. there may be no change.
Those that do want to change and implement, will not only see an improvement in their throughput, but will improve overall network performance.
Fairness? The improvement according to the proposed scheme does not come at the expense of degradation somewhere else. The performance of the entire network is improved.
Todor Cooklev
Todor Cooklev

23. May 2006
doc.: IEEE /0657r3
July 2006
Conclusions
AIFS is a critical parameter for delivering priority services and QoS in wireless home environment.
The proposed random-AIFSN has benefit to all traffic classes – throughput increases and delay decreases
While the suggested scheme is optional, and all current implementations comply with it as a special case.
Todor Cooklev
Todor Cooklev