Presentation is loading. Please wait.

Presentation is loading. Please wait.

Doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 1 VDCF Simulations Greg Chesson, Aman Singla,

Similar presentations


Presentation on theme: "Doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 1 VDCF Simulations Greg Chesson, Aman Singla,"— Presentation transcript:

1 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 1 VDCF Simulations Greg Chesson, greg@atheros.com Aman Singla, aman@atheros.com

2 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 2 Overview Notation Load(a,b,c) means there are a highest-priority stations, b at the next priority, c at the next priority, etc. May be shown as L(a,b,c,d). CWmin(a,b,c) describes the assignment of CWmins to Traffic Categories where a is assigned to the highest priority, b at the next, etc. May be shown as W(a,b,c,d). QIFS(a,b,c), or Q(a,b,c), captures the setting of the QIFS[i] values by TC. Simulation Software –Public version of the Berkeley NS2 software suite –Contact authors for details Results –Usually 3 graphs per simulation BW - An instantaneous bandwidth slot for one or more flows LAT – a per-frame latency plot over the lifetime of the run Lat-dist – latency distribution (i.e P% of the packets had latency less than x) Scenarios –Usually fixed topology, varying load –Stations are added or removed usually on 3-sec intervals

3 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 3 Scenarios 1.Basic differentiation demonstration and comparison with legacy DCF 2.Load(4,2,10) – phones, 2 videos, 10 background stations 3.Load(4,4,8) – constant demand from stations, uses same W/Q settings as scenario (2), demonstrates robust behavior of mechanism 4.Extreme differentiation: 8 active TCs

4 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 4 Scenario 1 A simple scenario meant only to compare with DCF and to observe the effects of CWmin and QIFS controls. Load(2,4): two high-priority stations and 4 low-priority stations, all stations with backlogged queues (always ready to transmit) using UDP datagrams. Topology: an AP is sinking all traffic from the 6 stations. Simulation Runs (3 plots for each run: bw, lat, lat-dist): a)DCF only - for comparison b)W(15,31) Q(0,0)- uses CWmin but not QIFS c)W(15,15) Q(0,1)- uses QIFS but not CWmin

5 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 5 1 a bw Per-flow Dropped packet counts for IFQ/MAC IFQ means dropped from software queue MAC means retry count exceeded

6 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 6 1 a lat

7 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 7 1 a lat-dist

8 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 8 1 b bw

9 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 9 1 b lat

10 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 10 1 b lat-dist

11 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 11 1 c bw

12 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 12 1 c lat-dist

13 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 13 Scenario 2 Load(4,2,10) –4 phone-like CBR flows (100 Kbit/s with 120B frames) Bidirectional flows: 2 Stations plus AP, 4 flows, each STA is source+sink –2 video-like CBR flows: 1 at 3 Mbit/s, 1 at 8 Mbit/s –10 background flows: 5 UDP and 5 TCP –All other flows sink to AP unless specified otherwise –New station added every 3 sec. –Stations removed in same order after peak load is reached Runs a.W(15,15,31) Q(0,2,7) b.W(15,15,31) Q(0,0,7) c.W(15,15,31) Q(0,2,7)- video flows sourced at AP

14 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 14 2 a bw phones videos background Remove loads

15 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 15 2 a lat phone video statistics

16 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 16 2 a lat-dist

17 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 17 2 a vdcf-background vs dcf-background

18 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 18 2 b bw Background flows not shown

19 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 19 2 b lat

20 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 20 2 b lat-dist

21 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 21 2 c bw video sourced at AP Background flows not shown

22 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 22 2 c lat video sourced at AP Lower latency than 2 b lat. Fewer phone-video collisions than 2-b.

23 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 23 2 c lat-dist video sourced at AP

24 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 24 Scenario 3 Load(4,4,8) – differentiation test using same W/Q as scenario 2 –4 high priority stations, 4 middle priority, 8 low priority –All stations sending with backlogged queues to AP –Stations are sequenced on at 3-sec intervals, high-priority stations first –Stations are sequenced off after peak load is attained Demonstrates good differentiation without needing to adjust W/Q a.Uses W(15,15,31) Q(0,2,7)- same as scenario 2 b.Uses W(15,15,15) Q(0,1,2)- different recipe

25 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 25 3 a bw

26 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 26 3 b bw

27 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 27 3 a lat

28 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 28 3 b lat

29 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 29 3 a lat-dist

30 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 30 3 b lat-dist

31 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 31 Scenario 4 8 streams, 8 TCs Backlogged queues

32 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 32 4 bw

33 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 33 4 lat

34 doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 34 4 lat-dist


Download ppt "Doc.: IEEE 802.11-01/133 Submission March 2001 G. Chesson, A. Singa - Atheros Slide 1 VDCF Simulations Greg Chesson, Aman Singla,"

Similar presentations


Ads by Google