Improving Capacity of VoIP in IEEE 802.11 Networks

Post on 01-Feb-2016

42 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

Improving Capacity of VoIP in IEEE 802.11 Networks. Takehiro Kawata Sangho Shin Andrea G. Forte Henning Schulzrinne. Need for support of many simultaneous calls Stadium Concert. Motivation. VoIP in Wireless Networks WIFI Phone, VoIP Client App for PDAs. Limited Capacity - PowerPoint PPT Presentation

Transcript

Improving Capacity of VoIP Improving Capacity of VoIP in IEEE 802.11 Networksin IEEE 802.11 Networks

Takehiro KawataSangho Shin

Andrea G. ForteHenning Schulzrinne

2Sangho ShinNov 16 2005

MotivationMotivation VoIP in Wireless Networks

WIFI Phone, VoIP Client App for PDAs

Limited Capacity 802.11a/b/g 11 Mb/s ~ 54 Mb/s Limited APs due to interferences

Need for support of many simultaneous calls Stadium Concert

3Sangho ShinNov 16 2005

OutlineOutline

Medium Access Control (MAC) in IEEE 802.11 LANs

Theoretical Capacity of VoIP in IEEE 802.11 LANs

Enhanced MAC Protocol : DPCF Simulation and results Conclusions

4Sangho ShinNov 16 2005

MAC Protocol in IEEE 802.11MAC Protocol in IEEE 802.11 Distributed Coordination Function (DCF)

Default MAC protocol

Contention Window

Busy Medium

DIFS DIFS

CSMA/CA

Backoff Next frame

Defer Access Slot

Backoff SlotsRTS

CTS

DATA

ACK

SIFS SIFS SIFS DIFS CW

Virtual Carrier Sense Mechanism (Four way handshake)

RTS: Request To Send, CTS: Clear To Send

DIFS

5Sangho ShinNov 16 2005

MAC Protocol in IEEE 802.11MAC Protocol in IEEE 802.11 Point Coordination Function (PCF)

For real time traffics Supports QoS (rudimentary) Optional, almost not implemented commercially

Beacon D1+poll

U1+ACK

D2+Ack+poll

U2+ACK

CF-End

SIFS SIFS SIFS SIFS SIFS

Contention Free Period (CFP)

Contention Period (CP)

Contention Free Repetition Interval (Super Frame)

poll

Null

SIFSDCF

PIFS

PIFS < SIFS < DIFS

6Sangho ShinNov 16 2005

1

Theoretical Capacity of Theoretical Capacity of VoIPVoIP

1

23

N

1Time

Packetization Interval

1

1

1

1 1

2

2

2

2

2

2

3

3

3

3

3

3

N

N

N

N

N

N….

….

….

….

….

….

7Sangho ShinNov 16 2005

1

Theoretical Capacity of Theoretical Capacity of VoIPVoIP

1

23

N

1Time

Packetization Interval

1

1

1

1 N+1

2

2

2

2

2

2

3

3

3

3

3

3

N

N

N

N

N

N….

….

….

….

….

….

N+1

N+1

N+1

N+1

N+1

N+1

1

Delay Delay Delay

• Constant Bit Rate (CBR)• No Silence Suppression

8Sangho ShinNov 16 2005

Theoretical Capacity of Theoretical Capacity of VoIPVoIP DCF

1

1Time

Packetization Interval

1

1

1

1 1

2

2

2

2

2

2

3

3

3

3

3

3

N

N

N

N

N

N….

….

….

….

….

….

Voice SIFS ACKBackOffDIFS

PLCP MAC IP RTP PayloadUDP

N calls = Packtization Interval / ( 2 * TTotal)

TTotoal = TDIFS+Tbackoff+Tvoice+TSIFS+TACK

Average backoff time = CW/2 * Slot time

TTotal = Time for sending a voice packet

9Sangho ShinNov 16 2005

Theoretical Capacity of Theoretical Capacity of VoIPVoIP DCF (IEEE

802.11b)  Time (us)

Length (B)

PLCP Preamble 144.00 18

PLCP Header 48.00 6

PLCP 192.00 24

MAC Header+CRC 26.18 36

IP   20

UDP   8

RTP   12

Voice (G711 + 20 ms pkt Int) 116.36 160

ACK 10.18 14

SIFS 10.00  DIFS 50.00  Contention Window (CW) 31 (number)  Slot time 20.00  

10Sangho ShinNov 16 2005

Theoretical Capacity of Theoretical Capacity of VoIPVoIP DCF (IEEE 802.11b)

3.6

5.7

9.0

10.7

0

2

4

6

8

10

12

1 2 3 4 5 6 7 8 9 10 11Bit Rate (Mb/s)

Num

ber

of c

alls

11Sangho ShinNov 16 2005

Theoretical Capacity of Theoretical Capacity of VoIPVoIP PCF

1

1Time

Contention Free Period

2

2

3

3

N

N….

….

VoiceSIFS

PLCP MAC IP RTP PayloadUDP

N calls = (CFP -TBeacon-TSIFS-TCF-End-TPIFS)/(2 * TTotal)

TTotoal = TSIFS+Tvoice

CP

B C

1

1

2

2

3

3

N

N….

….

B C

1

1

2

2B

Packtization Interval = CFP IntervalCP = 0

PIFS

12Sangho ShinNov 16 2005

Theoretical Capacity of Theoretical Capacity of VoIPVoIP PCF (IEEE 802.11b)

4.4

8.1

17.0

24.9

0

5

10

15

20

25

30

1 2 3 4 5 6 7 8 9 10 11Bit Rate (Mb/s)

Num

ber

of c

alls

13Sangho ShinNov 16 2005

Theoretical Capacity of VoIPTheoretical Capacity of VoIP

VoiceSIFS

Voice SIFS ACKBackOffDIFS

PCF

DCF

DCF vs PCF

PCF has so good performance also in VoIP traffic (w Silence Suppression) ?Ncalls = 25 /active ratio = 25 / 0.4 = 62 calls ??

3.65.7

9.0

4.37

8.06

17.04

24.9

10.7

0

5

10

15

20

25

30

1 2 3 4 5 6 7 8 9 10 11Bit Rate (Mb/s)

Num

ber

of c

alls

14Sangho ShinNov 16 2005

Problems of PCFProblems of PCF Waste of polls

VoIP traffic with Silence Suppression

1

poll

1

poll

Data

1

poll poll

NullData

poll

Null

poll

Null

1

ACK

1

ACK

1

ACK

Talking Period Mutual Silence Period Listening Period

Data

1

poll

1

poll

2Null

poll poll

1

poll

1

poll

2Null

poll poll

1

poll

1

poll

2Null

poll poll

1

poll

1

poll

2Null

poll poll

10 ms

AP

Various packetization intervalsNode 1 : 10 ms, Node 2 : 20 ms, AP: 10 ms PCF Intervals

15Sangho ShinNov 16 2005

Problems of PCFProblems of PCF Synchronization between polls and Data

poll poll poll

Null

CFP CPpoll

Null

poll

App

MAC

Node side

Polling time

Packet generation time

1MAC

2 3 4

5 6 7

CFP CP

AP side

1 2 3 4

CFP CP

NullNull

5 6 7

Polling time

16Sangho ShinNov 16 2005

Dynamic PCFDynamic PCF Classification of traffics

Real-time traffic (VoIP) Use PCF, also CF

Best effort traffic Use only CF

Give higher priority to real-time traffics

Best effort

1

poll

3

poll

5

poll

1 3 5 7 9

7

poll

9

poll

VoIP

MAC

1

CFP CP

Polling List 3 8

1

poll

3

poll

8

poll

VoIP

17Sangho ShinNov 16 2005

Dynamic PCFDynamic PCF

Dynamic Polling List Store only “active” nodes

MAC

CFP CP

Polling List 1 2 3 4

MAC

CFP CP

5678

1 3 8

65

Null NullACK

1 2 3 4

ACK

7 8

7

ACK

567

Queue

CFP CP

567

CFP CP

PCF

Polling List 1 3 8DPCF

5

1 3

1 2 3 4

Null

6 7 8

65 7

Null ACKACKACK

8

1 3 8

pollpoll poll

65

ACKACK

7

ACK

1 3 8

pollpoll poll

65

ACKACK

7

ACK

18Sangho ShinNov 16 2005

Dynamic Polling List Adding a node to a polling list

As soon as AP detects VoIP packet in CP

Removing a node from a polling list After AP gets 2 consecutive Null packets

5

CFP CP

MAC

1 3 8Polling List

CFP CP

MAC

Polling List 1 3 8 5

Dynamic PCFDynamic PCF

51

poll

3

poll

8

poll

1

poll

3

poll

8

poll

1

poll

3

poll

8

poll

5

poll

1

poll

3

poll

8

poll

1

poll

3

poll

5

poll

8

poll

1

poll

3

poll

5

poll

8

poll

19Sangho ShinNov 16 2005

Dynamic PCFDynamic PCF Dynamic CFP Interval and More data field

Use the biggest packetization interval as a CFP interval.

Set “more data field” when there are more than two packets to send.

Solution to the various packetization intervals problem

1

poll

Node 1 : 10 ms, Node 2 : 20 ms, AP: 20 ms PCF Intervals

2

poll poll

20 ms

1

poll

+more

21

poll

1

poll

1

poll

1

poll

2

poll

+more

2

poll

1

poll

+more

AP

20Sangho ShinNov 16 2005

poll poll

Dynamic PCFDynamic PCF More data field

Solution to the synchronization problem

poll

Null

CFP CPpoll

App

MAC

Node side

pollpoll

+more

Fail to send

21Sangho ShinNov 16 2005

Dynamic PCFDynamic PCF Synchronization problem in DPCF

MAC

AP side

1 2

7 8

CFP CP

Polling time

5

MAC1 2

7 8

CFP CP

Polling time

5

PCF

DPCF

22Sangho ShinNov 16 2005

App

MAC

DPCF

CFP CP

Dynamic PCFDynamic PCF Solution to the Synchronization problem

Allow VoIP packets to be sent in CP only when there are more than two VoIP packets in queue

poll poll

Null

poll pollCFP CP

poll poll

App

MAC

PCF

poll

poll

+more

poll pollpoll

23Sangho ShinNov 16 2005

SimulationsSimulations QualNet Simulators

Commercial simulator, evaluation available

Better physical model than NS2 Easy graphical + text interface

Topology : Ethernet to Wireless

AP

MN1

MN2

MN3

MN4

Router

CN1

CN2

CN3

CN4

24Sangho ShinNov 16 2005

SimulationsSimulations VoIP traffic model

ITU-T P59

Parameter Duration (s)

Rate (%)

Talk-spurt 1.004 38.53

Pause 1.587 61.47

Double-Talk 0.228 6.59

Mutual Silence

0.508 22.48

0.51.0 0.23 0.30.9

1.5

Duration (s)

Rate (%)

0.9 37.5

1.5 62.5

0 0

0.3 25.0

Our Model

25Sangho ShinNov 16 2005

SimulationsSimulations Deciding the capacity of VoIP

Threshold : 50 ms

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0 5 10 15 20 25 30 35 40

Number of VoIP

End

-to

-En

d D

ela

y /

Jitte

r (s

ec)

Delay uplink (90%)Jitter uplink (90%)Delay downlink (90%)Jitter downlink (90%)

26Sangho ShinNov 16 2005

Simulation ResultsSimulation Results

0

5

10

15

20

25

30

35

40

45

0 1 2 3 4 5 6 7 8 9 10 11 12

Transmission Rate (Mbps)

Nu

mb

er

of

Vo

IP N

od

es

DPCF

PCF

DCF

SimulationCalculation

42

34

33

27Sangho ShinNov 16 2005

Simulation ResultsSimulation Results Delay and throughput with FTP traffics

DCF (33 nodes)

21.3

215.5 219.2 230.6

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

0 1 2 3

Number of FTP Sessions

Th

rou

gh

pu

t (k

bp

s)

0

50

100

150

200

250

300

350

400

450

500

En

d-t

o-E

nd

De

lay

(ms)

FTP Throughput (download)

VoIP Throughput

VoIP Delay (90%)

28Sangho ShinNov 16 2005

Simulation ResultsSimulation Results Delay and throughput with FTP traffics

PCF (33 nodes)

228.8

168.6

109.1

42.0

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

0 1 2 3

Number of FTP Sessions

Th

rou

gh

pu

t (k

bp

s)

0

50

100

150

200

250

300

350

400

450

500

En

d-t

o-E

nd

De

lay

(ms)

FTP Throughput (download)

VoIP Throughput

VoIP Delay (90%)

29Sangho ShinNov 16 2005

Simulation ResultsSimulation Results Delay and throughput with FTP traffics

DPCF (33 nodes)

25.323.122.320.1

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

0 1 2 3

Number of FTP Sessions

Th

rou

gh

pu

t (k

bp

s)

0

50

100

150

200

250

300

350

400

450

500

En

d-t

o-E

nd

De

lay

(ms)

FTP Throughput (download)

VoIP Throughput

VoIP Delay (90%)

30Sangho ShinNov 16 2005

228.8

168.6

109.1

42.0

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

0 1 2 3

Number of FTP Sessions

Th

rou

gh

pu

t (k

bp

s)

0

50

100

150

200

250

300

350

400

450

500

En

d-t

o-E

nd

De

lay

(ms)

FTP Throughput (download)

VoIP Throughput

VoIP Delay (90%)

Simulation ResultsSimulation ResultsDCF 33 nodes PCF 33 nodes

DPCF 33 nodes

21.3

215.5 219.2 230.6

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

0 1 2 3

Number of FTP Sessions

Th

rou

gh

pu

t (k

bp

s)

0

50

100

150

200

250

300

350

400

450

500

En

d-t

o-E

nd

De

lay

(ms)

FTP Throughput (download)

VoIP Throughput

VoIP Delay (90%)

25.323.122.320.1

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

0 1 2 3

Number of FTP Sessions

Th

rou

gh

pu

t (k

bp

s)

0

50

100

150

200

250

300

350

400

450

500

En

d-t

o-E

nd

De

lay

(ms)

FTP Throughput (download)

VoIP Throughput

VoIP Delay (90%)

31Sangho ShinNov 16 2005

ConclusionsConclusions Estimated capacity of VoIP

theoretically and with simulation. Proposed Dynamic PCF.

Dynamic Polling List More data field Synchronization

Improved VoIP capacity by 25%. With FTP traffics, DPCF has lower

delay and higher throughput.

Thank you

http://www.cs.columbia.edu/IRT/wireless

33Sangho ShinNov 16 2005

Simulation ResultsSimulation Results

0

5

10

15

20

25

0 1 2 3 4 5 6 7 8 9 10 11 12

Transmission Rate (Mbps)

Nu

mb

er

of

Vo

IP N

od

es

PCF

DCF

SimulationCalculation

top related