Presentation is loading. Please wait.

Presentation is loading. Please wait.

Doc.: IEEE 802.11-02/214r1 Submission March 2002 Wentink, IntersilSlide 1 A simpler and better EDCF Menzo Wentink Ron Brockmann Maarten Hoeben Intersil.

Similar presentations


Presentation on theme: "Doc.: IEEE 802.11-02/214r1 Submission March 2002 Wentink, IntersilSlide 1 A simpler and better EDCF Menzo Wentink Ron Brockmann Maarten Hoeben Intersil."— Presentation transcript:

1 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 1 A simpler and better EDCF Menzo Wentink Ron Brockmann Maarten Hoeben Intersil

2 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 2 Related letter ballot comments

3 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 3 Categories of LB comments EDCF justificationEDCF justification –Complexity –Performance – can it do the job EDCF requests for enhancementsEDCF requests for enhancements No solution for IBSS operationNo solution for IBSS operation No proven solution for overlapNo proven solution for overlap

4 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 4 Line of thinking

5 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 5 Line of thinking (1) In the current EDCF, IFS and CW are used for latency and bandwidth differentiationIn the current EDCF, IFS and CW are used for latency and bandwidth differentiation However: the CW space is limited as a differentiator for both latency and bandwidth, and AIFS is complexHowever: the CW space is limited as a differentiator for both latency and bandwidth, and AIFS is complex Instead, the CFB length per priority is a more effective bandwidth differentiatorInstead, the CFB length per priority is a more effective bandwidth differentiator Separate latency and bandwidth differentiationSeparate latency and bandwidth differentiation –CW for latency (CWmin/CWmax) –CFB size for bandwidth Fixed parameter values!Fixed parameter values!

6 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 6 Line of thinking (2) So for EDCF, we propose to differentiate three parameters:So for EDCF, we propose to differentiate three parameters: –CWmin, CWmax, CFB length

7 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 7 Applications - bandwidth vs latency Bandwidth Latency CFB length CWmin/CWmax excellent effort video best effort HDTV interactive voice 10 ms100 ms 0.5 ms 3 ms 0/115/10237/15 64 kbps 25 Mbps

8 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 8 RTSThreshold = 300 Bytes.RTSThreshold = 300 Bytes. Suggested EDCF default settings priorityCWminCWmax CFB limit 0 (be) frame exch. 1 (ee) ms 2 (it) frame exch. 3 – spare (vi) ms 5 (vo) 010.5ms 6 – spare – spare ---

9 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 9 Simulations

10 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 10 Simulations (1) 24 Mbps PHY rate, unless noted otherwise24 Mbps PHY rate, unless noted otherwise g slots (20 usec) are used, unless noted otherwise802.11g slots (20 usec) are used, unless noted otherwise RTSThreshold = 300 BytesRTSThreshold = 300 Bytes Simulations are performed with Network SimulatorSimulations are performed with Network Simulator

11 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 11 Simulations (2) Traffic types in each scenarioTraffic types in each scenario –Voice: UDP, 290 Byte packets, 30 pps (G.711) –Video: UDP, 2300 Byte packets, increasing load (+1 Mbps per second) –Data: TCP, 1500 Byte packets All flows are upstreamAll flows are upstream Note that the video latency and voice latency plots have a different scaleNote that the video latency and voice latency plots have a different scale

12 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide nodes 7 voice streams 1 video stream 7 data streams (16:7/1/7)

13 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 13 7 voice / 1 video (36) / 7 data / 16 nodes voice: CWmin=0, CWmax=1, CFB=0.5ms (24 Mbps) video: CWmin=7, CWmax=15, CFB=3.0ms (36 Mbps) data: CWmin=15, CWmax=1023 (24 Mbps) applied video load (Mbps) throughput per class data throughput video throughput - increases linearly with the offered load total throughout voice bandwidth – unaffected by other load video throughput capped when the applied video load exceeds a maximum level 25 Mbps

14 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 14 7 voice / 1 video (36) / 7 data / 16 nodes voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 applied video load (Mbps) time fractions data fraction video fraction total useful time fraction voice fraction 100%

15 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 15 7 voice / 1 video (36) / 7 data / 16 nodes voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 applied video load (Mbps) video throughput is capped when the applied video load exceeds a maximum level the video throughput rises linearly with the offered video load 25 Mbps

16 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 16 7 voice / 1 video (36) / 7 data / 16 nodes voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 applied video load (Mbps) video latency the video latency stays low over a very wide range sharp increase when the offered load exceeds the maximum possible 50ms

17 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 17 7 voice / 1 video (36) / 7 data / 16 nodes voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 applied video load (Mbps) voice latency the voice latency stays very low, even under heavy overload conditions 10ms

18 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 18 Explanation of the EDCF parameters

19 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 19 Why does CFB differentiation work EDCF assigns TXOPs independent of the TXOP lengthEDCF assigns TXOPs independent of the TXOP length Longer CFBs results in higher usage per TXOP, for same CW, thus higher bandwidthLonger CFBs results in higher usage per TXOP, for same CW, thus higher bandwidth CFBs reduce network contention and increase the efficiency!CFBs reduce network contention and increase the efficiency! Effect of CFBs increases when overall traffic load increases, because the CFBs are more fully usedEffect of CFBs increases when overall traffic load increases, because the CFBs are more fully used

20 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 20 CFBs decrease latency and jitter CFBs help to reduce network contention and collisionsCFBs help to reduce network contention and collisions –jitter is reduced CFBs allow to use relatively high CWmin values for high-bandwidth priority streamsCFBs allow to use relatively high CWmin values for high-bandwidth priority streams Lower CWmin values can be used for low- latency applications like voiceLower CWmin values can be used for low- latency applications like voice

21 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 21 Another advantage of CFBs Higher PHY rates result in increasingly inefficient EDCF operationHigher PHY rates result in increasingly inefficient EDCF operation –IFS and slot-time overhead becomes larger compared to the length of frames Time-limited CFBs easily scale up to higher bitrates by aggregating more frames into a single CFBTime-limited CFBs easily scale up to higher bitrates by aggregating more frames into a single CFB

22 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 22 CWmin = 0? Ideal for transmitters with low traffic intensityIdeal for transmitters with low traffic intensity Exponential backoff ensures CW increaseExponential backoff ensures CW increase Very often, the medium is free at DIFSVery often, the medium is free at DIFS –nodes with interrupted backoff never transmit at DIFS, because minimum remaining backoff after deferral is 1 slot --> DIFS+1 is the soonest!!. Fixed CWmin=0 automatically adapts to the traffic load (more interrupted backoff)Fixed CWmin=0 automatically adapts to the traffic load (more interrupted backoff)

23 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 23 CWmax Worst-case latency and jitter are determined by CWmaxWorst-case latency and jitter are determined by CWmax –after few collisions the current CW can become very large, even when started with a low CWmin –latency insensitive traffic should defer to latency sensitive traffic in case of collisions. –CWmax differentiation helps bounding jitter for low-latency streams The effect automatically adapts to traffic load, based on fixed CWmax settingsThe effect automatically adapts to traffic load, based on fixed CWmax settings

24 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 24 RTS/CTS Protects no-Ack burstsProtects no-Ack bursts Resolves hidden nodes situationsResolves hidden nodes situations Reduces the cost of collisionsReduces the cost of collisions The effect scales with the traffic load, based on a fixed RTS thresholdThe effect scales with the traffic load, based on a fixed RTS threshold

25 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 25 Implementation is extremely simple CWmin: existingCWmin: existing CWmax: existingCWmax: existing CFBs: similar to fragment burstingCFBs: similar to fragment bursting RTS/CTS: existingRTS/CTS: existing No IFS differentiationNo IFS differentiation No new frame formatsNo new frame formats No dynamic parameter updatesNo dynamic parameter updates Simple, but the functionality is great!Simple, but the functionality is great!

26 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 26 Conclusions

27 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 27 Conclusions (1) Excellent support for voice under any scenario (< 10 ms delay)Excellent support for voice under any scenario (< 10 ms delay) Excellent support for video (<< 50 ms delay)Excellent support for video (<< 50 ms delay) Excellent bandwidth and latency differentiationExcellent bandwidth and latency differentiation Very high channel efficiencyVery high channel efficiency Scales transparantly to higher rate PHYsScales transparantly to higher rate PHYs Works in overlap and non-overlap scenariosWorks in overlap and non-overlap scenarios Works for IBSS and infrastructureWorks for IBSS and infrastructure Supports unicast, multicast and no-ACK (FEC)Supports unicast, multicast and no-ACK (FEC) Works upstream, downstream and sidestreamWorks upstream, downstream and sidestream

28 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 28 Conclusions (2) Requires no changes to current MAC framesRequires no changes to current MAC frames Simple - Minimal set of mechanismsSimple - Minimal set of mechanisms Constant parameters - no parameter-tuning intelligence needed in APConstant parameters - no parameter-tuning intelligence needed in AP –No possible oscillations between different implementations in overlap situations Implicit accommodation of bursty trafficImplicit accommodation of bursty traffic Legacy STAs can participate in QoS network, and receive QoS flowsLegacy STAs can participate in QoS network, and receive QoS flows –No need for legacy upgrade Simple to Analyze/Simulate/ImplementSimple to Analyze/Simulate/Implement Predictable behaviorPredictable behavior Works with hidden nodesWorks with hidden nodes

29 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 29 Explanation of the classesExplanation of the classes –be: Best Effort –ee: Excellent Effort –it: Interactive TCP –vi: Video –vo: Voice Default settings (2)

30 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 30 Motions

31 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 31 Contention Free Bursting This motion will be entertained on behalf of the HCF Ad Hoc groupThis motion will be entertained on behalf of the HCF Ad Hoc group

32 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 32 Motion 1 Instruct the Editor to incorporate changes to the TGe draft, such that EDCF differentiation is limited to CWmin, CWmax and the TXOP limit per priority.Instruct the Editor to incorporate changes to the TGe draft, such that EDCF differentiation is limited to CWmin, CWmax and the TXOP limit per priority.

33 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 33 Instruct the Editor to specify the mandatory default values for the following MIB variables:Instruct the Editor to specify the mandatory default values for the following MIB variables: –RTSThreshold = 300 Bytes. Motion 2 idot11CWmin[i]dot11CWmax[i]dot11CPCFBlimit[i] 0 (be) aCWminaCWmax0 1 (ee) aCWminaCWmax1.5ms 2 (it) (vi) ms 5 (vo) 010.5ms

34 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 34 Motion 3 Instruct the Editor to remove the QoS Parameter Set elementInstruct the Editor to remove the QoS Parameter Set element

35 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 35 More simulations

36 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 36 Simulations All flows are upstream (sidestream)All flows are upstream (sidestream) Traffic typesTraffic types –voice: UDP, 290 Byte packets, 30 pps (G.711) –video: UDP, 2300 Byte packets, increasing load (+1 Mbps per second) –data: TCP, 1500 Byte packets Note that the video and voice latency plots have a different scaleNote that the video and voice latency plots have a different scale NotationNotation –4:1/1/1 for 4 active nodes, 1 voice, 1 video, 1 background – for CWmin=7, CWmax=15, CFB size=3.0ms

37 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 37 4:1/1/

38 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 38 1 voice (24) / 1 video (36) / 1 data (24) / 4 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) throughput per class 25 Mbps

39 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 39 1 voice / 1 video (36) / 1 data / 4 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) time fractions 100%

40 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 40 1 voice / 1 video (36) / 1 data / 4 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) video throughput 25 Mbps

41 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 41 1 voice / 1 video (36) / 1 data / 4 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) video latency

42 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 42 1 voice / 1 video (36) / 1 data / 4 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) voice latency

43 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 43 18:7/3/

44 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 44 7 voice / 3 video (36) / 7 data / 19 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) throughput per class 25 Mbps

45 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 45 7 voice / 3 video (36) / 7 data / 19 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) time fractions per class 100%

46 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 46 7 voice / 3 video (36) / 7 data / 19 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) video throughput 12 Mbps

47 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 47 7 voice / 3 video (36) / 7 data / 19 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) video latency

48 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 48 7 voice / 3 video (36) / 7 data / 19 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) voice latency

49 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 49 16:7/1/

50 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 50 7 voice / 1 video (36) / 7 data / 16 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) throughput per class 25 Mbps

51 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 51 7 voice / 1 video (36) / 7 data / 16 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) time fraction per class 100%

52 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 52 7 voice / 1 video (36) / 7 data / 16 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) video throughput 20 Mbps

53 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 53 7 voice / 1 video (36) / 7 data / 16 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) video latency

54 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 54 7 voice / 1 video (36) / 7 data / 16 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) voice latency

55 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 55 18:7/3/

56 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 56 7 voice / 3 video (36) / 7 data / 18 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) throughput per class 25 Mbps

57 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 57 7 voice / 3 video (36) / 7 data / 18 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) time fraction per class 100%

58 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 58 7 voice / 3 video (36) / 7 data / 18 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) video throughput

59 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 59 7 voice / 3 video (36) / 7 data / 18 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) video latency

60 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 60 7 voice / 3 video (36) / 7 data / 18 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=15, CWmax=15, CFB=4.5ms data: CWmin=15, CWmax=1023, CFB=0 video load (Mbps) voice latency

61 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 61 8:3/1/ CCK RTS/CTS

62 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 62 3 voice / 1 video (36) / 3 data / 8 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.5ms data: CWmin=15, CWmax=1023 video load (Mbps) throughput per class 25 Mbps

63 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 63 3 voice / 1 video (36) / 3 data / 8 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.5ms data: CWmin=15, CWmax=1023 video load (Mbps) time fraction per class

64 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 64 3 voice / 1 video (36) / 3 data / 8 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.5ms data: CWmin=15, CWmax=1023 video load (Mbps) video throughput 25 Mbps

65 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 65 3 voice / 1 video (36) / 3 data / 8 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.5ms data: CWmin=15, CWmax=1023 video load (Mbps) video latency

66 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 66 3 voice / 1 video (36) / 3 data / 8 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.5ms data: CWmin=15, CWmax=1023 video load (Mbps) voice latency

67 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 67 10:3/3/ a PHY

68 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 68 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) throughput per class 25 Mbps

69 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 69 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) time fraction per class

70 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 70 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video throughput 25 Mbps

71 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 71 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video latency

72 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 72 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) voice latency

73 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 73 10:5/1/ % PHY frame loss (+ collisions)

74 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 74 5 voice / 1 video (36) / 5 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) throughput per class 25 Mbps

75 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 75 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) time fraction per class

76 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 76 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video throughput 25 Mbps

77 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 77 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video latency

78 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 78 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) voice latency

79 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 79 10:5/1/ % PHY frame loss (+ collisions)

80 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 80 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) throughput per class 25 Mbps

81 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 81 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) time fraction per class

82 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 82 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video throughput 25 Mbps

83 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 83 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video latency

84 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 84 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) voice latency

85 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 85 10:5/1/ hidden video node

86 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 86 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) throughput per class 25 Mbps

87 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 87 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) time fraction per class

88 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 88 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video throughput 25 Mbps

89 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 89 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video latency

90 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 90 3 voice / 3 video (36) / 3 data / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) voice latency

91 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 91 8:3/1/ Best effort at 12Mbps

92 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 92 3 voice / 3 video (36) / 3 data (12) / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) throughput per class 25 Mbps

93 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 93 3 voice / 3 video (36) / 3 data (12) / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) time fraction per class

94 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 94 3 voice / 3 video (36) / 3 data (12) / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video throughput 25 Mbps

95 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 95 3 voice / 3 video (36) / 3 data (12) / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) video latency

96 doc.: IEEE /214r1 Submission March 2002 Wentink, IntersilSlide 96 3 voice / 3 video (36) / 3 data (12) / 10 voice: CWmin=0, CWmax=1, CFB=0.5ms video: CWmin=7, CWmax=15, CFB=3.0ms data: CWmin=15, CWmax=1023 video load (Mbps) voice latency


Download ppt "Doc.: IEEE 802.11-02/214r1 Submission March 2002 Wentink, IntersilSlide 1 A simpler and better EDCF Menzo Wentink Ron Brockmann Maarten Hoeben Intersil."

Similar presentations


Ads by Google