Checking Improving TCP and Voice
Performance using a 802.11e testbed
David Malone, Ian Dangerfield
Anthony Ng & Doug Leith, Hamilton Institute
20 January 2006
1
802.11b TCP
Performance
12 uploads
6 up, 6 down
12 TCP Uploads to AP for 180 sec
6 Uploads and 6 Downloads for 180 sec
1200
1200
1000
800
Throughput(kbits/sec)
Throughput(kbits/sec)
1000
600
400
200
0
800
600
400
200
1
2
3
4
5
6
7
8
STA id
9
10
11
0
12
1
2
3
4
5
Have proposed fixes, want to test in practice.
2
6
7
8
STA id
9
10
11
12
802.11(e) 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.
Ideally produces even distribution of packet
transmissions.
In 11e have multiple queues. Each has own CWmin ,
DIFS(aka AIFS) and can have TXOP.
3
Why use a testbed?
• Can we believe ns?
Bugs: aCCATime, virtual collisions.
• Can we believe the standard?
• Can we believe models?
• What are the practical issues?
4
Testbed setup
Number of identical stations (Linux) connection to AP
(Linux hostap).
1× AP
Dell GX 280
2.8Ghz P4
12× STA
Soekris net4801
266Mhz 586
WLAN
D-Link DWL-G520
Atheros AR5212
Cards have external antenna, PCI interface, Madwifi
driver with local patches for 11e parameter setting.
MGEN and iperf used for traffic generation.
5
6
Practical Issue:
Calibration
UDP up
TCP up
11 STAs Upload To AP UDP
11 STAs Upload To AP
700
1500
600
Throughput(kbits/sec)
Throughput(kbits/sec)
500
400
300
1000
500
200
100
0
1
2
3
4
5
6
STA id
7
8
9
10
0
11
Small changes until well behaved.
7
1
2
3
4
5
6
STA id
7
8
9
10
11
Validation
Relative Throughput against TXOP parameter
Relative Throughput against AIFS parameter
5.5
30
Analytical Model
Experimental Results
5
25
4
Relative Throughput
Relative Throughput
4.5
3.5
3
2.5
20
15
10
2
5
1.5
1
0
1000
2000
3000
4000
TXOP in unit of us
5000
6000
7000
Relative Throughput against CWmin
20
18
Analytical Model
Experimental Results
16
Relative Throughput
14
12
10
8
6
4
2
0
−3
−2
−1
0
1
2
3
CWmin Shift (0 corresponds to the default CWmin 32)
4
0
0
5
10
15
AIFS in unit of 20us timeslot
20
25
Measure relative performance of two saturated flows while varying TXOP, AIFS and
CWmin . Compare to
well-known models.
8
Proposed settings
AIFS
CWmin
(slots)
AP
TXOP
(packets)
Upload ACKs
0
4
1
Download data
4
32
nd
wireless
Download ACKs
0
32
1
station
Upload data
4
32
1
Derived using analytical modeling and ns.
Will they work in practice?
9
Before
12 uploads
6 up, 6 down
12 TCP Uploads to AP for 180 sec
6 Uploads and 6 Downloads for 180 sec
1200
1200
1000
800
Throughput(kbits/sec)
Throughput(kbits/sec)
1000
600
400
200
0
800
600
400
200
1
2
3
4
5
6
7
8
STA id
9
10
11
0
12
10
1
2
3
4
5
6
7
8
STA id
9
10
11
12
After
12 uploads
6 up, 6 down
12 TCP Uploads to AP for 180 sec
6 Uploads and 6 Downloads
600
700
600
500
Throughput(kbits/sec)
Throughput(kbits/sec)
500
400
300
200
400
300
200
100
0
100
1
2
3
4
5
6
7
8
STA id
9
10
11
0
12
11
1
2
3
4
5
6
7
8
STA id
9
10
11
12
Voice
• Voice is quite different to TCP.
• Has a loss and delay requirement.
• Low rate vs. high rate.
• Aim to protect voice from saturated sources.
• AIFS is the obvious parameter.
• (simulation says 4, model says 6 to be safe).
12
Unprioritised Voice
Throughput
Delay
70
30000
60
Average Delay (seconds x 10-6)
25000
50
Throughput (kbps)
Unpriotised Station
Delay bound for queue stability
40
30
20
20000
15000
10000
5000
10
Unpriotised Station
Ideal Throughput
90% Throughput
0
0
2
4
6
0
8
10
12
14
16
18
Number of competing stations
0
2
4
6
8
10
12
Number of competing stations
13
14
16
18
Measuring Delay
• Want to measure one-way MAC delay.
• Tried NTP for a laugh.
• Tried simultaneously observable broadcasts.
• Transmission not complete until MAC ACK.
• Hardware supports interrupt after ACK.
14
Validation
1600
Median delay (seconds x 10-6)
1400
1200
1000
800
600
400
Long preamble
Fit line: 11.0012Mbps x + 484.628us
Short preamble
Fit line: 11.0016Mbps x + 292.565us
200
0
0
200
400
600
800
1000
1200
Packet size (bytes, excluding headers)
15
1400
4000
16000
1400
1400
3500
14000
600
1500
400
1000
Mean delay (us)
800
2000
Throughput x delay (bytes)
Mean delay (us)
1000
2500
1200
12000
0
500
0
1000 1500 2000 2500 3000 3500 4000 4500
35000
Mean delay (us)
1000
25000
800
20000
600
15000
400
10000
200
5000
Throughput x Mean delay (Station 2)
Mean delay (Station 2)
300
400
CWmin value second station
500
0
600
Throughput x delay (bytes)
1200
30000
200
2
4
6
0
8
10
AIFS value of second station
1400
100
200
Throughput x Mean delay (Station 2)
Mean delay (Station 2)
0
40000
0
400
0
TXOP value for station 1
0
600
6000
2000
Throughput x Mean delay (Station 1)
Mean delay (Station 1)
0
800
8000
4000
200
500
1000
10000
Throughput x delay (bytes)
1200
3000
Measure relative performance of two saturated
flows
while
varying
TXOP, AIFS and CWmin .
Check Throughput *
delay has expected value.
16
AIFS Impact
Throughput
Delay
70
30000
60
Average Delay (seconds x 10-6)
25000
50
Throughput (kbps)
Unpriotised Station
Competitors AIFS = 4
Competitors AIFS = 6
Delay bound for queue stability
40
30
20
Unpriotised Station
Competitors AIFS = 4
Competitors AIFS = 6
Ideal Throughput
90% Throughput
10
0
0
2
4
6
20000
15000
10000
5000
0
8
10
12
14
16
18
Number of competing stations
0
2
4
6
8
10
12
Number of competing stations
17
14
16
18
Delay Distribution
AIFS 0
AIFS 6
100
Percentage of packets (less than associated delay)
Percentage of packets (less than associated delay)
100
80
60
40
1 competing station
6 competing stations
12 competing stations
18 competing stations
Interpacket time
20
0
0
20000
40000
60000
80000
80
60
40
1 competing station
6 competing stations
12 competing stations
18 competing stations
Interpacket time
20
0
100000
Delay (seconds x 10-6)
0
20000
40000
60000
Delay (seconds x 10-6)
18
80000
100000
Autocorrelation
1
Unpriotized, 1 competing station
Unpriotized, 6 competing stations
Unpriotized, 12 competing stations
Unpriotized, 18 competing stations
AIFS = 4, 1 competing station
AIFS = 4, 6 competing stations
AIFS = 4, 12 competing stations
AIFS = 4, 18 competing stations
AIFS = 6, 1 competing station
AIFS = 6, 6 competing stations
AIFS = 6, 12 competing stations
AIFS = 6, 18 competing stations
0.8
autocorrelation
0.6
0.4
0.2
0
-0.2
0
5
10
15
number of packets
19
20
25
30
Conclusions
• Small operational testbed.
• Hardware seems to behave as expected.
• Radio issues can be amplified by other issues.
• 11e can be used to combat MAC/TCP issues.
• 11e can be used to help voice out.
• Further looking at mixed voice/data and voic only
networks.
20