Towards Backoff Range-Based Service Differentiation over IEEE 802.11 Wireless LAN Networks

1
Towards Backoff Range-Based Service
Differentiation over IEEE 802.11 Wireless LAN
Networks
SCW
2
Abstract
Best effort ??????????traffic?????
3
Abstract
DCF 802.11 MAC ??????,?????traffic??????(??priorit
y)?????,??????QoS???
? ?
??QoS
EDCF (enhanced DCF) AEDCF (adaptive EDCF) AF-EDCF
(adaptive fair EDCF) SCW(sliding contention
window)
4
Introduction-EDCF

• ??DCF??? ??? priority ???????????
  (?8? priority queues)

• DCF???queue????EDCF 8 ??????queue,??TC(traffic
  class)
• ??TC???virtual DCF station(?????queue),?????stati
  on???8?virtual DCF station?????????frame????

5
Introduction-EDCF(Cont.)

• AIFS i (arbitration interframe space)
• ??DCF?DIFS(AIFSgtDIFS)
• ??TC?????AIFS,??????TC?TC?AIFS??

3 mechanisms

• CWmax i maxmun contention window value

• CWmin i minimum contention window value

• Allow each TCs window to have different behavior

6
Introduction-EDCF(Cont.)

• In DCF
• In EDCF

CWnew 2 CWold
CWnewi (CWoldi 1) PFi
??TC????PF( Persistence factor ) ?
?????,PF???,??????????TC?????????
7
Introduction-AEDCF
EDCF ??TC?PF??????,?????scalability problem
? ?
AEDCF(adaptive EDCF)?PF???????,????????(collision
rate)????CW?
8
Introduction-AF-EDCF
EDCF?AEDCF ????????,?????????????,???????????,??
????TC???????bandwidth,????????????
? ?
AF-EDCF(adaptive fair EDCF)? channel load
????,????????TC???
9
Introduction-SCW
EDCF?AEDCF?AF-EDCF
SCW(sliding contention window)

• ??? node ?????????,????????,??????? traffic
  ????????? TC? flows,??? flows ?????????
• backoff time ??????,???? QoS guarantees, fairness
  and bandwidth efficiency

• ?? flows ???????
• ?? TC ???? CW range,?? backoff counter is
  selected dependent on the type of traffic being
  transmitted

? ?
10
Introduction-SCW (Cont.)
SCW i is associated with each TC i
11
Introduction-SCW (Cont.)

• CW i LB SCW is lower bound
• CW i UP SCW is upper bound
• The lower and upper bounds delimit the interval
  from which TCis flow select a random backoff
  value

12
Introduction-SCW (Cont.)
The bounds of the window change as the window
slides, but stay within the interval CW imin,
CW imax
13
Introduction-SCW (Cont.)

• Sliding factor,?? PF i
• TC i priority ??,SF i ??
• stride,?? CW up or down

14
Introduction-SCW (Cont.)

• ????

When loss are low packets are transmitted
successfully
When loss are high
15
Introduction-SCW (Cont.)
SLIDING ALGORITHM
16
Introduction-SCW (Cont.)
SLIDING ALGORITHM
Loss rate(drop rate)
17
Introduction-SCW (Cont.)
SLIDING ALGORITHM
A threshold value for the maximum tolerated loss
rate for TC i
18
Introduction-SCW (Cont.)
SLIDING ALGORITHM
If the loss rate is too high,gives the TC higher
priority and reducing the loss rate
19
Introduction-SCW (Cont.)
SLIDING ALGORITHM
If the loss rate is low,give more opportunity to
lower-priority flows
20
Introduction-SCW (Cont.)
SLIDING ALGORITHM
Does not require any QoS metric thresholds
21
Introduction-SCW (Cont.)
SLIDING ALGORITHM
Network load
22
Simulation model

• 10 wireless terminals(WT i , i 110)
• A single access point( AP )
• Each WT generates up to 3 different flows at a
  time
• High priority(HP)
• Medium priority(MP)
• Best effort(BE)

23
Simulation model-HP flows throughput
24
Simulation model-HP flows throughput
Mean throughput SCW72.27 kb/s EDCA70.37
kb/s AEDF69.88 kb/s
25
Simulation model-HP flows throughput
SCW
7 kb/s
The throughput is not good due to high intraclass
contention provoked by a too narrow backoff range
for HP flows
26
Simulation model-HP flows throughput
AEDCF EDCA
The throughput is relatively good due to the use
of a scenario that the bit rate peaks do not
occur at the same time
27
Simulation model-HP flows delay
28
Simulation model-HP flows delay
AEDCF and EDCA suffer from higher queuing delays
at the MAC level
29
Simulation model-MP flows throughput
30
Simulation model-MP flows throughput
Too narrow backoff range provokes EDCAs MP flows
high intra-TC contention and a serious drop in
throughput
31
Simulation model-MP flows delay
32
Simulation model-MP flows delay
EDCAs MP flows experience occasional
degradations that have consequences for the delay
33
Simulation model-BE flows throughput
34
Simulation model-BE flows throughput
SCW achieve better overall network utilization
during this period
35
Simulation model-BE flows throughput
SCW outperforms AEDCF with 100 kb/s in network
utilization gain
36
Simulation model-BE flows throughput
Compared to SCW, AEDCF has excessive throughput
oscillations because it uses a per-flow collision
rate to adjust the backoff range
37
Simulation model-overall network throughput
38
Simulation model-overall network throughput
EDCA suffers a devastating drop in network
utilization during the period
39
Conclusions and future works

• Simulations show that the SCW protocol performs
  better than EDCA ADECF
• Reduces oscillation
• Increases fairness
• Currently, SCW class-specific MAC parameters must
  be carefully adjusted, based on the WLAN
  deployment scenario and predicted traffic.
• Hence, future work should focus on finding
  mechanisms for dynamically adapting these
  parameters to particular network scenarios.