An Enhanced EDD QoS Scheduler for IEEE 802.11e WLAN - Advances in Computational Science and Engineering - page 56

Fig. 6. Throughput in scenario 1 (CBR trac)
Figure 7 shows the mean delay time in scenario 1. In the figure, we can see
that the average delay for the reference scheduler is the highest and the delays
for SETT-EDD and the proposed algorithm are the same. In the simulation, the
beacon interval and the packet generation interval are set to 100ms and 30ms,
respectively. Since 20ms is the greatest common divisor, the reference scheduler,
allocated SI as 20ms. Packets arrived in the previous SI need to wait maximum
20ms even if there is only one STA. This is because the reference scheduler is
not work-conserving. However, the other two schedulers are work conserving.
Thus, if there are packets to transmit, they try to transmit the packet instead
of idling the transmitter. This explains why the average delays of our scheduler
and SETT-EDD are smaller than that of the reference scheduler.
Fig. 7. Average delay in scenario 1 (CBR trac)
Figure 8 and 9 shows the throughput and the the packet loss ratio in scenario
2 in which MPEG-4 video trac was used. In figure 9, we can see that packet
drop occurs in the reference scheduler even when the number of STAs is two.
This is because the reference scheduler could not handle large burst. SETT-EDD