1

Mona Habib & Nirmala BulusuMona Habib & Nirmala Bulusu

Improving QoS of VoIP over Wireless Networks

(IQ-VW)

Improving QoS of VoIP over Wireless Networks

(IQ-VW)

CS522 – 12/09/2002

2

AgendaAgenda

� Voice over IP (VoIP): Why?

� VoIP Protocols: H.323 and SIP

� Quality of Service (QoS)

� Wireless Networks

� Testbed Configuration

� Testing Scenarios

� QoS Test Results

� Comments

� Voice over IP (VoIP): Why?

� VoIP Protocols: H.323 and SIP

� Quality of Service (QoS)

� Wireless Networks

� Testbed Configuration

� Testing Scenarios

� QoS Test Results

� Comments

3

� Reduce toll costs for long-distance telephone calls

� Helps consolidate separate voice and data networks for cost-effectiveness and bandwidth utilization.

� Provides features not available in traditional voice telephony, such as video conferencing and simultaneous data transmission (e.g. whiteboard) for true multimedia communications.

� Provides integration between data and telephony applications for business -- “click to talk” on a web site for ordering or customer support.

� Reduce toll costs for long-distance telephone calls

� Helps consolidate separate voice and data networks for cost-effectiveness and bandwidth utilization.

� Provides features not available in traditional voice telephony, such as video conferencing and simultaneous data transmission (e.g. whiteboard) for true multimedia communications.

� Provides integration between data and telephony applications for business -- “click to talk” on a web site for ordering or customer support.

Why Voice over IP?Why Voice over IP?

4

IP NetworkIP NetworkGatekeeperGatekeeper

ISDNISDN

WirelessWireless

PSTNPSTN

GatewayGateway

TerminalTerminal

TerminalTerminal

FaxFaxPBXPBX

Enterprise ClassEnterprise ClassCarrier ClassCarrier Class

Enterprise NetworkEnterprise Network

GatewayGateway

MCUMCU

SS7 NetworkSS7 Network

Voice over IP Network ComponentsVoice over IP Network Components

5

Voice over IP - The Standards BattleVoice over IP - The Standards Battle

• H.323- Primary standard for enterprise networks- Supported in many carrier networks

• SIP - Session Initiation Protocol- Common for IP phones and PCs - Gaining popularity as signaling protocol due to its

versatility

• H.323- Primary standard for enterprise networks- Supported in many carrier networks

• SIP - Session Initiation Protocol- Common for IP phones and PCs - Gaining popularity as signaling protocol due to its

versatility

6

Link & Physical Layer

IP

UDP TCP

RTP

Voice CodecG.711, 723,

729, etc. RTCP H.225RAS

H.225 Call

SignalingH.245

Audio Application Terminal Control & Management

H.323 Protocol StackH.323 Protocol Stack

7

Link & Physical Layer

IP

UDP TCP

RTP

Voice CodecG.711, 723,

729, etc. RTCP SIP SDP

Audio Application Terminal Control & Management

SIP Protocol StackSIP Protocol Stack

8

Codec: Speech to DataCodec: Speech to Data

Transforms between Speech (analog) to data (digital) Bandwidth

101011110

Algorithm G.723.1 G.729 G.728 G.726

Rate (Kbs) 5.3 - 6.3 8 16 32

Complexity Highest High Lower Low

Compare with 64Kbs end to endCompare with 64Kbs end to end

9

Quality of ServiceQuality of Service

• Offered Service• Reachability• Availability• Reliability• Price

• Offered Service• Reachability• Availability• Reliability• Price

Service QualityService Quality Voice QualityVoice Quality

Traditional PSTN• Level• Delay• Echo• Clarity:

• Intelligibility• Noise• Fading• Cross talk

Traditional PSTN• Level• Delay• Echo• Clarity:

• Intelligibility• Noise• Fading• Cross talk

In addition in IP Networks• Delay• Delay-Jitter• Clarity:

• Packet Loss• Bandwidth• Compression

In addition in IP Networks• Delay• Delay-Jitter• Clarity:

• Packet Loss• Bandwidth• Compression

GatewayGateway IP NetworkIP NetworkPSTN NetworkPSTN Network H.323/SIP TerminalH.323/SIP TerminalPhonePhone

10

Wireless NetworksWireless Networks

802.11aOperates in the 5 GHz frequency band

Supports bandwidths up to 54 MB, range of 150+ feet

Has 12 data channels

Uses Orthogonal Frequency Division Multiplexing (OFDM)

Performs at short distances

Incompatible with 802.11b

802.11aOperates in the 5 GHz frequency band

Supports bandwidths up to 54 MB, range of 150+ feet

Has 12 data channels

Uses Orthogonal Frequency Division Multiplexing (OFDM)

Performs at short distances

Incompatible with 802.11b

802.11b� Operates in the 2.4 GHz

frequency band

� Supports bandwidths up to 11 MB , range of 150+ feet

� Has 3 data channels

� Uses Direct Sequence Spread Spectrum modulation (DSSS)

� Handles long distances better than 802.11a

802.11b� Operates in the 2.4 GHz

frequency band

� Supports bandwidths up to 11 MB , range of 150+ feet

� Has 3 data channels

� Uses Direct Sequence Spread Spectrum modulation (DSSS)

� Handles long distances better than 802.11a

802.11 is an IEEE standard for wireless LANs

802.11a and 802.11b are two variants of the standard

Most recent variant: 802.11g (compatible with 802.11b)

802.11 is an IEEE standard for wireless LANs

802.11a and 802.11b are two variants of the standard

Most recent variant: 802.11g (compatible with 802.11b)

11

Wireless Network SecurityWireless Network Security

� Vulnerabilities:

� Unauthorized user access

� Eavesdropping (network can be tapped using a sniffer)

� Authentication: EAP (Extensible Authentication Protocol)� EAP interacts with a Remote Authentication Dial-In User

Service (RADIUS) server to provide authentication for wireless client devices.

� Encryption: WEP (Wired Equivalent Privacy)� Scrambles the communication between the access point and

client devices to keep the communication private.

� Both the access point and client devices use the same WEP key to encrypt and decrypt radio signals.

� Vulnerabilities:

� Unauthorized user access

� Eavesdropping (network can be tapped using a sniffer)

� Authentication: EAP (Extensible Authentication Protocol)� EAP interacts with a Remote Authentication Dial-In User

Service (RADIUS) server to provide authentication for wireless client devices.

� Encryption: WEP (Wired Equivalent Privacy)� Scrambles the communication between the access point and

client devices to keep the communication private.

� Both the access point and client devices use the same WEP key to encrypt and decrypt radio signals.

12

Wireless Network ConfigurationWireless Network Configuration

13

QoS Testbed – HW ConfigurationQoS Testbed – HW Configuration

Gatekeepercalvin.uccs.edu

Gatekeepercalvin.uccs.edu

Ethernet Clientwait.uccs.edu

Ethernet Clientwait.uccs.edu

Ethernet Clientwind.uccs.eduEthernet Clientwind.uccs.edu

Wireless Client(DHCP)

Wireless Client(DHCP)

Wireless Client(DHCP)

Wireless Client(DHCP)

Lab #3Lab #3

Lab #2Lab #2 Lab #1Lab #1

Cisco AironetAccess PointCisco AironetAccess PointRADIUS Server

vinci.uccs.eduRADIUS Servervinci.uccs.edu

14

QoS Testbed – SW ConfigurationQoS Testbed – SW Configuration

� Public Domain Software

� Gatekeeper: Vovida Open Communication Application Library (VOCAL)

� VOCAL SIP to H.323 Converter: SIPH323CSGW

� Clients: MSN Messenger 4.6 (allows use of communication services other than .Net Passport)

� Network Analyzer: Ethereal

� Other Software:

� QoS analysis tools provided by Daniel Hertrich

� Voice over Misconfigured Internet Telephones (VOMIT)

� Wavfix.c: Program to create WAVE file header. Used to replay captured voice data

� Public Domain Software

� Gatekeeper: Vovida Open Communication Application Library (VOCAL)

� VOCAL SIP to H.323 Converter: SIPH323CSGW

� Clients: MSN Messenger 4.6 (allows use of communication services other than .Net Passport)

� Network Analyzer: Ethereal

� Other Software:

� QoS analysis tools provided by Daniel Hertrich

� Voice over Misconfigured Internet Telephones (VOMIT)

� Wavfix.c: Program to create WAVE file header. Used to replay captured voice data

15

QoS Testing ScenariosQoS Testing Scenarios

� Ethernet to Ethernet

� Ethernet to Wireless

� Ethernet to 802.11a

� Ethernet to 802.11b

� Ethernet to 802.11b + Wireless security

� Wireless to Wireless

� 802.11a to 802.11a

� 802.11b to 802.11b

� 802.11b to 802.11b + Wireless security

� Ten test runs per scenario. Sound files include speech (male and female) and music.

� Ethernet to Ethernet

� Ethernet to Wireless

� Ethernet to 802.11a

� Ethernet to 802.11b

� Ethernet to 802.11b + Wireless security

� Wireless to Wireless

� 802.11a to 802.11a

� 802.11b to 802.11b

� 802.11b to 802.11b + Wireless security

� Ten test runs per scenario. Sound files include speech (male and female) and music.

16

QoS Test ResultsSample Inter-packet Delay Graph

QoS Test ResultsSample Inter-packet Delay Graph

802.11b - 802.11b

0

4

8

12

16

20

24

28

32

36

40

0 10 20 30 40 50 60

Tm e ( s e c )

17

QoS Test ResultsSample Jitter Time Graph

QoS Test ResultsSample Jitter Time Graph

802.11b - 802.11b

-0.4

-0.36

-0.32

-0.28

-0.24

-0.2

-0.16

-0.12

-0.08

-0.04

0

0.04

0.08

0.12

0.16

0.2

0.24

0.28

0.32

0.36

0.4

0 10 20 30 40 50 60

Ti m e ( s e c )

18

QoS Test Results: Average Jitter per CallEthernet to Wireless (802.11a vs. 802.11b)QoS Test Results: Average Jitter per CallEthernet to Wireless (802.11a vs. 802.11b)

Ethernet to Wireless802.11a vs. 802.11b

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

1 2 3 4 5 6 7 8 9 10

C a l l N u m b e r s

Et he rne t 802.11a 802.11b

Distance from AP ~15-20 ft. Excellent signal strength with both 802.11a and 802.11b. 802.11a performed better than 802.11b.

19

QoS Test Results: Average Jitter per CallEthernet to Wireless (with and without WEP)

QoS Test Results: Average Jitter per CallEthernet to Wireless (with and without WEP)

Ethernet to Wireless802.11b vs. 802.11b + security

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

1 2 3 4 5 6 7 8 9 10

C a l l N u m b e rs

Et he rne t 802.11b 802.11b+S ec

More research/testing needed to validate these results. Tests performed on different days (maybe different network load).

20

QoS Test Results: Average Jitter per CallWireless to Wireless (802.11a vs. 802.11b)QoS Test Results: Average Jitter per CallWireless to Wireless (802.11a vs. 802.11b)

Wireless to Wireless802.11a vs. 802.11b

0

100

200

300

400

500

600

1 2 3 4 5 6 7 8 9 10

C a l l N u m b e r s

Et herne t 802.11a 802.11b

Each client’s distance from AP ~15-20 ft. Peer-to-peer ~30-40 ft. Poor signal strength for 802.11a, excellent for 802.11b.

21

QoS Test Results: Average Jitter per CallWireless to Wireless (with and without WEP)

QoS Test Results: Average Jitter per CallWireless to Wireless (with and without WEP)

Wireless to Wireless802.11b vs. 802.11b +security

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

1 2 3 4 5 6 7 8 9 10

C a l l N u m b e rs

Et he rne t 802.11b 802.11b+S ec

No significant difference in results. Need to investigate further to check at which point packets are captured by Winpcap.

22

QoS Test Results: Loss of DataQoS Test Results: Loss of Data

� No loss of data observed during all test runs (except the 802.11a to 802.11a test).

� Good subjective assessment of QoS (user listening to the received sound). Clear sound with no interruptions, with the exception of an initial delay.

� Poor signal strength during 802.11a to 802.11a test (20-40% on both ends).

� High data loss rate observed (for e.g., transmitted 196 out of 1434 packets). Loss rate of ~86-93%.

� Extremely poor sound quality (unintelligible, broken, …)

� Packets lost at the sender’s end, as seen by Ethereal captured data. This needs to be investigated further as it affects interpretation of the results.

� No loss of data observed during all test runs (except the 802.11a to 802.11a test).

� Good subjective assessment of QoS (user listening to the received sound). Clear sound with no interruptions, with the exception of an initial delay.

� Poor signal strength during 802.11a to 802.11a test (20-40% on both ends).

� High data loss rate observed (for e.g., transmitted 196 out of 1434 packets). Loss rate of ~86-93%.

� Extremely poor sound quality (unintelligible, broken, …)

� Packets lost at the sender’s end, as seen by Ethereal captured data. This needs to be investigated further as it affects interpretation of the results.

23

QoS Test Results: ObservationsQoS Test Results: Observations

� Smooth sound quality for all test but 802.11a to 802.11a test, despite existing inter-packet delays and jitters.

� Replaying captured packets (after reassembling Wave file) reflected inconsistent delays, yet sound was clear at the receiving client.

� Sound quality improved by client’s handling of timings (e.g., using RTP to synchronize relative timings).

� Quality variations were most perceived in test #7, which had highest overlap of speech and music.

� Loss of data has the highest effect on QoS.

� Smooth sound quality for all test but 802.11a to 802.11a test, despite existing inter-packet delays and jitters.

� Replaying captured packets (after reassembling Wave file) reflected inconsistent delays, yet sound was clear at the receiving client.

� Sound quality improved by client’s handling of timings (e.g., using RTP to synchronize relative timings).

� Quality variations were most perceived in test #7, which had highest overlap of speech and music.

� Loss of data has the highest effect on QoS.

Let’s listen to a sample reassembled sound file … …

24

How to Improve QoS?How to Improve QoS?

Problem� Delay and jitter

� Packet loss due to congestion

Problem� Delay and jitter

� Packet loss due to congestion

Solution� Separate queues for time

sensitive traffic� RTP

� More bandwidth� Resource Reservation Protocol

(RSVP)� Differentiated Service (DiffServ)� Multi-Protocol Label Switching

(MPLS)� RFC 2597 and RFC 2598

Solution� Separate queues for time

sensitive traffic� RTP

� More bandwidth� Resource Reservation Protocol

(RSVP)� Differentiated Service (DiffServ)� Multi-Protocol Label Switching

(MPLS)� RFC 2597 and RFC 2598

25

Future Research/TestsFuture Research/Tests� Inject background traffic� Synchronize time on all testbed components and

calculate initial connection delay� Evaluate the effect of using different codecs� PC-to-Phone quality testing� Evaluate wireless network performance at different

distances from the access point� Evaluate wireless network performance using multiple

access points with overlapping coverage� Assess compatibility of 802.11 variants� Evaluate existing QoS solutions (e.g., RSVP)� Evaluate QoS of VoIP using H.323 clients� Detect transmission sampling rate for replay based on

timestamps of the captured packets� Evaluate QoS of VoIP using a PDA client (might require

porting a SIP client to a PDA)

� Inject background traffic� Synchronize time on all testbed components and

calculate initial connection delay� Evaluate the effect of using different codecs� PC-to-Phone quality testing� Evaluate wireless network performance at different

distances from the access point� Evaluate wireless network performance using multiple

access points with overlapping coverage� Assess compatibility of 802.11 variants� Evaluate existing QoS solutions (e.g., RSVP)� Evaluate QoS of VoIP using H.323 clients� Detect transmission sampling rate for replay based on

timestamps of the captured packets� Evaluate QoS of VoIP using a PDA client (might require

porting a SIP client to a PDA)

26

ReferencesReferences

� Collins, Daniel (2001). Carrier Grade Voice over IP, McGraw-Hill.

� Ferguson, Paul and Geoff Huston (1998). Quality of Service, Wiley Computer Publishing.

� Douskalis, Bill (2000). IP Telephony: The Integration of Robust VoIP Services, Prentice Hall PTR.

� Keey, David G., Cullen Jennings, and Luan Dang (2002). Practical VoIP using VOCAL, O’Reilly Network.

� Gast, Matthew (2002). 802.11 Wireless Networks: The Definitive Guide, O’Reilly Network.

� Hertrich, Daniel et al. (2001). “Evaluating QoS for Voice over IP in Wireless LANs”, Technical Report, Telecommunication Networks Group.

� Useful links: VoIP-WLAN-QoS Useful Links

� Collins, Daniel (2001). Carrier Grade Voice over IP, McGraw-Hill.

� Ferguson, Paul and Geoff Huston (1998). Quality of Service, Wiley Computer Publishing.

� Douskalis, Bill (2000). IP Telephony: The Integration of Robust VoIP Services, Prentice Hall PTR.

� Keey, David G., Cullen Jennings, and Luan Dang (2002). Practical VoIP using VOCAL, O’Reilly Network.

� Gast, Matthew (2002). 802.11 Wireless Networks: The Definitive Guide, O’Reilly Network.

� Hertrich, Daniel et al. (2001). “Evaluating QoS for Voice over IP in Wireless LANs”, Technical Report, Telecommunication Networks Group.

� Useful links: VoIP-WLAN-QoS Useful Links