802'11e EDCA performance analysis

1
802.11e EDCA performance analysis

• Yoram Haddad
• JCT(Jerusalem College of Technology), Israel
• Gwendal Legrand
• ENST(Telecom Paris), France
• JDIR2007

2
Agenda

• Issue
• Overview of the 802.11e EDCA
• Existing models
• Our model
• Throughput derivation
• Results (non official)
• Future works

3
Issue

• Im a network administrator and I have to know
  how many hot spots I need, to deploy a wireless
  network in our university ...
• Solution 1 simply divide the nominal throughput
  per cell by the offered traffic per station !
• Wrong answer because we have to take into
  account errors that occur in channel, all
  protocols overhead, etc
• For example lets take a 802.11b cell (11Mbps)
  and compute how many VoIP users (128 kbps) can it
  support ? ? 85 users !!!
• Experiments have been made and have shown that it
  cant support more than 6 users for an acceptable
  QoS

4
Solution 2,3

• Solution 2 derive the overhead introduced by the
  protocols along the OSI model
• Partly wrong answer because when 2 stations
  transmit simultaneously, a collision occurs !
• For this purpose we need a model that represents
  the access mechanism to the network (MAC layer)

5
Overview of the 802.11e EDCA

• 802.11e is the extension of the 802.11 standard
  (Wi-Fi) which introduce QoS
• QoS support is provided by the introduction of
  access categories (ACs) and multiple independent
  backoff entities.

6
EDCA parameters per AC

• AC prioritization is realized by setting
  different values for the AC parameters. The most
  important of which are
• Arbitrary inter-frame space number (AIFSN) The
  minimum time interval between the wireless medium
  becoming idle and the start of transmission of a
  frame.
• Contention Window (CW) A random number is drawn
  from this interval, or window, for the backoff
  mechanism.
• TXOP Limit The maximum duration for which a QSTA
  can transmit after obtaining a TXOP.(not
  mandatory and can be disabled)

7
Existing analytical models for EDCA

• There are mainly two models of the DCF(802.11)
• Bianchis model based on markov chain
• Calis model backoff time based on geometric
  distribution
• Assumptions
• Each access category always has a packet to
  transmit Saturated Mode
• Ideal channel environment without errors
• Retry limit not taken into consideration
• From those two models , enhancement were made
  considering one of the above assumptions
• but there isnt one who consider all the
  assumptions in the same model
• We have to adapt it to EDCA

8
EDCA model our contribution

• We fully extend the original model proposed by
  Bianchi for DCF and Kong et al. for EDCA with
• Non-saturated mode (Merci M. Jacquet)
• Error prone channel
• The goal
• Derive the collision probability

9
4 dimension Markov Chain 1st dimension
10
other dimensions
11
The full markov chain
12
Throughput Calculation
ACi throughput Si can be expressed as
where P is the data packet payload size
13
Throughput

• Collision probability
• pi

14
Results M5, Pe0, q var
15
Results q1, BER010-5, M var
16
Future Works

• derive the delay
• Adapt the load factor q to specifics traffic
  scenario
• Study low level performance e.g. probability that
  specific slot times are occupied by MAC frame of
  given traffic

17
Merci

• Questions ?