An Enhanced EDD QoS Scheduler for IEEE 802.11e WLAN - Advances in Computational Science and Engineering - page 53

TXOP avg =max [ t pre − t ]

· ρ i

+ O, M i
R i

+ O (8)

R i

where t pre and t denote, respectively, the previous polled time for QST A i and
the present time. To transmit the packets remaining in the queue at previous SI,
it needs TXOP( TDr i ) is calculated exactly the same as ARROW. Hence, the
total TXOP is given as follows.

TXOP i = TXOP avg + TDr i

(9)

Figure 4 shows the method of TXOP allocation in the proposed algorithm.

Fig. 4. TXOP assignment with proposed algorithm

3.2 SI Allocation

When the algorithm determines SI, it considers two aspects. First one is that
the data generated in previous SI is delayed if the packets remain in queue as
illustrated in figure 5-(a). Another is that when QSTA does not consume given
TXOP, then it has longer idle time after data transmission as illustrated in figure
5-(b). In this case, the delay may become longer for subsequently arrived packets.
If the next service starts earlier, the delay can be reduced.
In the first case that there are the packets waiting in queue, it causes the delay
of the data arrived in previous SI as seen figure 5-(a). It is because the mSI which
the QSTAs can has a authority to transmit by mSI is fixed. However, as seen
in figure 5-(a), if mSI is decreased as much as

TDr

which is the time necessary
to transmit previously queued packets, we can avoid the delay by changing the
start of mSI. New mSI can be calculated as follows.

mSI new = mSI i − TDr cur

(10)

TDr i where denotes the value of current TXOP Duration request.

In the second case that QSTA does not consume given TXOP as illustrated in
figure 5-(b), the idle time after data transmission increases, which would lead to
increased delay for the packets arrived later. If next SI starts earlier so that the
idle time after data transmission remains the same, we could reduce the delay.
Hence our scheduler changes mSI as follows.

cur