Download presentation
Presentation is loading. Please wait.
1
EDCF / EPCF Comparisons
Month 2000 doc.: IEEE /xxx September 2002 EDCF / EPCF Comparisons Author: Matthew Sherman AT&T Labs - Research 180 Park Avenue Florham Park, NJ Date: September 8, 2002 Matthew Sherman, AT&T Labs - Research John Doe, His Company
2
Purpose of Document Quick comparison of EDCF with EPCF
Month 2000 doc.: IEEE /xxx September 2002 Purpose of Document Quick comparison of EDCF with EPCF Mostly for new comers Matthew Sherman, AT&T Labs - Research John Doe, His Company
3
Background Few comparisons of E-DCF and HCF
Month 2000 doc.: IEEE /xxx September 2002 Background Few comparisons of E-DCF and HCF AT&T has simulated both E-DCF and E-PCF E-DCF simulations – eg. 01/004 and 01/019 E-PCF simulations – eg. 02/303, 02/304, 02/305 E-PCF just like HCF except CF access only during CFP HCF should perform even better since less constrained Easier for AT&T to simulate E-PCF AT&T’s original interest before HCF created AT&T never provided head to head comparison First such comparisons by AT&T Matthew Sherman, AT&T Labs - Research John Doe, His Company
4
Simulation Scenario Same parameters as in 01/004
Month 2000 doc.: IEEE /xxx September 2002 Simulation Scenario Same parameters as in 01/004 01/004 did not use PHY Ran CC/RR/EDCF code with b PHY with b access parameters but no modeling of PHY overhead Some output statistics different as well Matthew Sherman, AT&T Labs - Research John Doe, His Company
5
Plots Collected Matched plots in 01/004 as closely as possible
September 2002 Plots Collected Matched plots in 01/004 as closely as possible Statistics not identical to CC/RR/EDCF model Stuck with Throughput (Goodput), Media Access Delay, and Data Dropped “Streams” here correlate to “Calls” in 01/004 Streams are renumbered (relative to Calls) to correlate with when streams start Matthew Sherman, AT&T Labs - Research
6
E-PCF sims of E-DCF Scenarios
September 2002 E-PCF sims of E-DCF Scenarios Configured code to run E-DCF Scenarios using E-PCF 20 msec Beacon Period (18 msec CFP) E-DCF simulations used 100 msec Beacon Period Standing Poll (no CC/RR) Addresses arranged according to priority Priority based scheduler no attempt at fairness Matthew Sherman, AT&T Labs - Research
7
Traffic loading Same EDCF parameters for classes as 00/004
September 2002 Traffic loading 2.5 Mbps LOW Traffic Category AIFS CW MIN CW MAX PF TOP PIFS 15 1023 1.5 MEDIUM DIFS 31 2.0 LOW DIFS +1 5.0 Mbps TOP 5.0 Mbps MEDIUM 2.5 Mbps LOW 30 90 150 Time (sec) Same EDCF parameters for classes as 00/004 Matthew Sherman, AT&T Labs - Research
8
E-PCF simulations of E-DCF scenarios
September 2002 E-PCF simulations of E-DCF scenarios Matthew Sherman, AT&T Labs - Research
9
Throughput E-DCF E-PCF September 2002
Matthew Sherman, AT&T Labs - Research
10
Media Access Delay E-DCF E-PCF September 2002
Matthew Sherman, AT&T Labs - Research
11
Media Access Delay (Zoomed)
September 2002 Media Access Delay (Zoomed) E-DCF E-PCF Matthew Sherman, AT&T Labs - Research
12
Data Dropped E-DCF E-PCF September 2002
Matthew Sherman, AT&T Labs - Research
13
September 2002 Data Analysis – E-PCF Dramatic performance differences between E-DCF and E-PCF High priority delays / drops much better for E-PCF Difference between acceptable / unacceptable Voice E-DCF much more sensitive to overload E-PCF drops less data in general No “fairness” between classes in E-PCF Could be fixed with more advanced scheduler HCF would be better still Matthew Sherman, AT&T Labs - Research
14
E-PCF simulations of E-DCF scenarios (No PHY)
September 2002 E-PCF simulations of E-DCF scenarios (No PHY) Matthew Sherman, AT&T Labs - Research
15
Throughput E-DCF E-PCF September 2002
Matthew Sherman, AT&T Labs - Research
16
Media Access Delay E-DCF E-PCF September 2002
Matthew Sherman, AT&T Labs - Research
17
Media Access Delay (Zoomed)
September 2002 Media Access Delay (Zoomed) E-DCF E-PCF Matthew Sherman, AT&T Labs - Research
18
Data Dropped E-DCF E-PCF September 2002
Matthew Sherman, AT&T Labs - Research
19
Data Analysis – E-PCF (no PHY)
September 2002 Data Analysis – E-PCF (no PHY) While the E-DCF performance degradation is not as sever as with the PHY it is still substantial All the comments identified for the simulations with a PHY would still apply Matthew Sherman, AT&T Labs - Research
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.