Verification of Common 802.11 MAC
Model Assumptions
David Malone, Ian Dangerfield and Doug Leith
Hamilton Institute, NUI Maynooth, Ireland.
6 April 2006
1
Motivation
• Lots of 802.11 modeling work.
• Modeling has been quite successful.
• However, models make simplifying assumptions.
• Theory to justify assumptions has proven hard.
• Can measurement help?
2
802.11 Summary
• After TX choose rand(0, CW − 1).
• Wait until medium idle for DIFS(50µs),
• While idle count down in slots (20µs).
• TX when counter gets to 0, ACK after SIFS (10µs).
• If ACK then CW = CWmin else CW∗ = 2.
3
Assumptions
Data
Ack
Collision
• Time is slotted.
• Stations transmit in a slot independently.
• Transmission/Collision probabilities fixed.
• 1 − p = (1 − τ )n−1 .
• S=
Ep nτ (1−τ )n−1
.
σ(1−τ )n +Ts nτ (1−τ )n−1 +(1−(1−τ )n −nτ (1−τ )n−1 )Tc
4
What Can We Test?
• We can measure delay, collisions and throughputs.
• Are cards close to 802.11?
• Is the timing good enough to slot time?
• Is collision probability really fixed?
• Does throughput formula work out?
5
Card timing
0.025
Fraction of Packets
0.02
0.015
0.01
0.005
0
900
1000
1100
1200
1300
Delay (seconds x 10-6)
6
1400
1500
1600
• 32 Uniform peaks, within jitter.
• Stefano, et al show not all cards so well behaved.
• Card trying to follow standard, timing plausible.
• Now look at collision probabilities.
• Use synthetic Poisson arrivals.
• Fixed number of stations.
7
Two Stations
0.06
0.05
Average P(col)
P(col on 1st tx)
P(col on 2nd tx)
1.0/32
1.0/64
Probability
0.04
0.03
0.02
0.01
0
100
200
300
400
500
600
Offered Load (per station, pps, 496B UDP payload)
8
700
800
Ten Stations
0.3
0.25
Probability
0.2
0.15
0.1
0.05
Average P(col)
P(col on 1st tx)
P(col on 2nd tx)
0
40
60
80
100
120
140
Offered Load (per station, pps, 486B UDP payload)
9
160
180
• Collision probability not constant.
• First stage dominates mean.
• Later stages closer for more stations.
• What about independence?
• Use transmit/collision/throughput relationship.
• Look at saturated traffic, vary stations.
10
Collisions vs Stations
0.35
0.3
Probability
0.25
0.2
0.15
0.1
0.05
P(col)
Predicted P(col)
0
2
4
6
8
Number of STA(s)
11
10
12
14
Throughput vs Stations
4000
3500
Throughput (Kbps)
3000
2500
2000
1500
1000
500
Throughput (measured)
Throughput (model)
Throughput (model + measured Ts/Tc)
Throughput (model + measured Ts/Tc/p)
0
2
4
6
8
Number of STA(s)
12
10
12
14
Conclusion
• n small — assumptions don’t hold.
• n large — assumptions closer.
• For small n maybe cancellation of errors.
• Can we now improve second order stats?
• Can we measure inter-station correlations?
• Can we get a better handle on slottedness?
13