SpeechTEK 2006 – Voice Over IP Tutorial Andrew Hunt, Ph.D. VP Engineering, Holly Connects Andrew Hunt, Ph.D. VP Engineering, Holly Connects

SpeechTEK 2006 – Voice Over IP TutorialSpeechTEK 2006 – Voice Over IP Tutorial

Andrew Hunt, Ph.D.VP Engineering, Holly Connects

Andrew Hunt, Ph.D.VP Engineering, Holly Connects

Welcome!

Who are you?

Andrew Hunt, Ph.D.VP of Engineering

Level 11, 301 George StreetSydney, NSW 2000, Australia

Tel +61 2 8207 8207Email: [email protected]: http://www.holly-connects.com/

4

DRAFT V03DRAFT V03 TimingTiming

8:30am Start

10:00-10:30am Coffee break

12:00-1:00pm Lunch

2:30-3:00pm Coffee break

4:30pm Close

5

DRAFT V03DRAFT V03 AgendaAgenda

1. Welcome & Introductions

2. Why Voice Over IP?

3. Brief History of Telephony

4. Digital Voice

5. Voice Over IP – Basics

6. VoIP Protocols

7. SIP: Session Initiation Protocol

8. RTP: Real-time Transport Protocol

9. Network Issues and Design

10. VoIP and Speech Recognition

11. VoIP and Mobile Telephony

12. Closing

6

DRAFT V03DRAFT V03 ObjectivesObjectives

Informative

Relevant

Interesting

Interactive

Why Voice Over IP?

8

DRAFT V03DRAFT V03 Module OverviewModule Overview

Why Voice Over IP?

What is it?

Cost

Functionality & flexibility

Mobility

9

DRAFT V03DRAFT V03 VoIP DefinitionVoIP Definition

Definitions

“Voice Over IP” is the use of the internet, intranets and other IP networks for the delivery of voice conversations

“Internet Protocol” (IP) is a protocol used for communicating data across a packet-switched network (specifically IPv4 or IPv6)

Many VoIP protocols exist: focus today on SIP and related protocols for use in speech recognition and IVR contexts

10

DRAFT V03DRAFT V03 Why VoIP?Why VoIP?

Cost

Global data traffic exceeded voice traffic in late 1990’s

Telco charges & revenue largely from voice traffic

Migration to shared networks

Single network for voice and data

Utilize spare capacity in many data networks

Inwards-out approach to migration

Traditional voice carriage costs to business being driven down by VoIP

Residential advantages

Free services: Skype etc.

Lower cost services: Vonage, Skype etc.

Regulatory and service capability issues are evolving

11


Functionality and Flexibility

Virtualization &mobility: move calls and agents locally, nationally, globally

Match traditional telephony functions

Transfers, voice mail, conferencing, redial, speed-dial, forwarding etc.

Integration with software and internet services

Availability notification (IM)

Instant messaging

Multi-media services: video, data files etc.

Options for traditional and VoIP telephony to co-exist and migrate steadily

12


Mobility

Make VoIP calls virtually anywhere in the world

Number portability: landline, mobile, internet

Synchronization of address books and contacts

Brief History of Telephony

14


Brief History of Telephony

Telephony switching

Basic concepts of telephony

Call establishment

Circuits

Switching

Migrating from analogue to digital

15

DRAFT V03DRAFT V03 Telephony: Switching and CircuitsTelephony: Switching and Circuits

Alexander Graham

Bell

Thomas Watson

16


1. Caller picks up

2. Caller “dials”

3. Callee phone rings

4. Callee answers

5. Circuit established

6. Conversation

7. Hang-up & tear-down

17


ExchangeExchange

ExchangeExchange

ExchangeExchange

ExchangeExchange

ExchangeExchange

18


PSTN: Public Switched Telephony Network

1. A-party picks up

2. A-party dials

3. B-party rings

4. B-party answers

5. Circuit established

6. Conversation

7. Hang-up & tear-down

19


Pre-Digital “Analogue” Era

1878: New Haven, Connecticut

World’s first commercial telephone exchange

Built from “carriage bolts, handles from teapot lids and bustle wire”

Cost: $40 including the furniture

1891: Topeka, Kansas

Almon Strowger, an undertaker, patented the Strowger switch

Automation of the telephone circuit switching by decadic pulses

1950 onwards: Crossbar switches

1964: “Dual Tone Multi-Frequency” (DTMF) introduced

20


Pre-Digital “Analogue” Era

Infrastructure

Local wiring to each phone

System of local, regional, national and international exchanges

Shared connections between exchanges

Call establishment

Numbering scheme (many iterations) Voice decadic pulsing DTMF

Mapping of numbering scheme to the exchanges (e.g. CAstle=22)

Call communication

Dedicated circuit per call

21


Digital Era

Digital started with the core telco networks

Efficiency on long-distance carriage

Efficiency of solid state switching technology

Migrated to local exchanges

What is digital voice?

Digital Voice

23


Digital Speech

Sampling: creating digital audio

CODECs: Compression and companding

Sharing channels: Time Division Multiplexing (TDM)

Standard digital data links: E1 & T1

24

DRAFT V03DRAFT V03 Telephony: Packet Switch NetworkTelephony: Packet Switch Network

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…


25

DRAFT V03DRAFT V03 Sampling TheorySampling Theory

26


27

DRAFT V03DRAFT V03

Sample frequency bandwidth

Sampling TheorySampling Theory

28

DRAFT V03DRAFT V03

Sam

ple

reso

lutio

n =

bits

n

oise

/err

orSampling TheorySampling Theory

29


Time (msec)0

Value0

1 292 573 824 1025 1176 1257 1268 1209 108

10 8911 6612 3913 914 -2015 -4916 -7517 -9718 -11419 -12420 -127… …

30

DRAFT V03DRAFT V03

Non

-line

ar s

ampl

ing

b

ette

r no

ise/

erro

rSampling Theory: CompandingSampling Theory: Companding

G.711 μ-law for North America and JapanG.711 A-law for Europe and rest of the world

31


Measure G.711 Telephony Compact Disc

Sampling rate 8,000 Hz 44,100 Hz

Frequency range Low – 3.5kHz (4kHz max) Lower – 22 kHz

Sample type8 bit A-law / μ-lawMono

16-bit linear PCMStereo

Signal-to-noise ratio < 70dB 96 dB

Data bandwidth 8 kByte/sec 176 kByte/sec

Perceived quality Telephony Great

32

DRAFT V03DRAFT V03 Digital TransportDigital Transport

• 10

• 52

• 87

• 92

• 10

2

• 85

• 49

• 10

• -1

8

• -4

8

• -6

0

• -1

2

• 10

2

• 85

• 49

• 10

• -1

8

• -4

8

• -6

0

• -1

2

• 10

• 52

• 87

• 92

• 1

0•

52

• 8

7•

92

• 1

02

• 8

5•

10

2•

85

• 4

9•

10

• -1

8•

-48

• -4

9•

10

• -1

8•

-48

• -6

0•

-12

• -6

0•

-12

• 1

0•

52

• 8

7•

92

TDM = TDM = Time Division Multiplex Time Division Multiplex

LatencyLatency PacketPacket

33


E1

World (ex. NA and Japan)

2.048 Mbit/s full duplex

32 time slots = 32 channels of 8-bit x 8 kHz

1 time slot reserved for framing

1 time slot is typically reserved for signalling

30 time slots for voice communications

E3 = 16 x E1 = 480 channels

T1

North America and Japan

1.536 MBit/s full duplex

24 time slots = 24 channels of 8-bit x 8kHz

34

DRAFT V03DRAFT V03 CompressionCompression

Compression of voice and audio

CODEC = COmpression - DECompression

Reduce bandwidth for the audio signal = more channels on the same transport

Lossless vs. lossy algorithms

Latency impact

CPU impact

Quality impact

Speech recognition impact

35


CODEC Characteristics Description

ITU-T G.711 Sample rate: 8kHzSample size: 8-bit A-law/μ-law Bandwidth: 64kbit/s

Standard telephony quality with “companding”

ITU-T G.726 Sample rate: 8kHzSample size: 2, 3, 4, 5-bitBandwidth: 16, 24, 32, 40 kbit/s

Adaptive Delta PCM (ADPCM). Supercedes G.721 & G.723

ITU-T G.728 Bandwidth: 16 kbit/sDelay: 5 samples, 0.625 ms

LDCELP = Low Delay Code Excited Linear Prediction

ITU-T G.729 Bandwidth: 8 or 6.4, 11.8 kbit/sDelay: 10 ms chunks

CS-ACELP = Conjugate-Structure Algebraic-Code-Excited Linear Prediction

ITU-T G.722 Sample rate: 16kHzSample size: 14-bit Bandwidth: 32-64kbit/s

Standard wideband speech ADPCM codec.Non-telephony CODEC

36


CODECs can increase channel capacity

1 E1/T1 channel = 64 kbit/s

= 1 channel of G.711

= 2 channels of G.726 @ 32 kbit/s


1 T1 = 24 channels

= 24 channels of G.711



BUT, CODECs can reduce voice quality

37


Advanced topics

Unreliable tone transmission on some CODECs

DTMF, Fax, Modem etc.

Use “out-of-band” communication (more later)

Silence suppression

Voice Activity Detection

Replace by simulated background noise

e.g. G.729 Annex B

Comfort noise generator (CNG)

Played when a communication channel fails temporarily

Usability: Reduces hang-up on temporary outages

38

DRAFT V03DRAFT V03

Voice Over IP – Basics

40


Voice Over IP Basics

Internet and intranet for voice communication

Challenges of the internet protocols

Quick guide to the internet protocols: TCP/IP, UDP

Applying internet protocols to session management & voice carriage

41

DRAFT V03DRAFT V03 Telephony: Packet Switch NetworkTelephony: Packet Switch Network

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…


42

DRAFT V03DRAFT V03 Voice Over Internet ProtocolVoice Over Internet Protocol

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

…0101100…

…0011100…

…1100110…

…1101100…

…0001100…

Intranet

OR

Internet

43

DRAFT V03DRAFT V03 ChallengesChallenges

Call establishment protocols

Connect A-party and B-party

Perform advanced telephony functions

Audio transport protocols

Get the packets from A-to-B / B-to-A on time

Latency

Packet loss

Jitter

Quality of service

44

DRAFT V03DRAFT V03 The ProblemThe Problem

The internet was not designed to carry real-time voice traffic!!!

(well almost)

45

DRAFT V03DRAFT V03 Internet Protocol Suite StackInternet Protocol Suite Stack

LinkLinkLayer 1

NetworkNetworkLayer 2

TransportTransportLayer 3

ApplicationApplicationLayer 4

Ethernet, Wi-Fi, ATM, Frame Relay…

IP (IPv4, IPv6)

TCP, UDP, SCTP, DCCP, IL, RUDP, …

DNS, FTP, HTTP, SMTP, SNMP, TELNET, SIP, RTP, H.323…

46

DRAFT V03DRAFT V03 Voice Over Internet ProtocolVoice Over Internet Protocol

Internet

ApplicationApplication


47


LinkLink

NetworkNetwork

TransportTransport


LinkLink

NetworkNetwork

TransportTransport

ApplicationApplicationPeer-to-peer connection

LinkLink

NetworkNetwork

LinkLink

NetworkNetwork

48


LinkLink

NetworkNetwork

TransportTransport


LinkLink

NetworkNetwork

TransportTransport


LinkLink

NetworkNetwork

LinkLink

NetworkNetwork

0101100

0101100 0101100

49


LinkLink

NetworkNetwork

TransportTransport


LinkLink

NetworkNetwork

TransportTransport


LinkLink

NetworkNetwork

LinkLink

NetworkNetwork

0101100

0101100 0101100

Packet Loss

50


TCP/IP = Transmission Control Protocol

Transport protocol

One of the core protocols of the Internet protocol suite: 75% of all traffic

Applications on networked hosts can create connections to one another using TCP for exchange of data or packets.

Guarantees reliable and in-order delivery of data from sender to receiver

Distinguishes data for multiple, concurrent applications on the same host

TCP supports many of the most popular application protocols including HTTP (Web), Email and SIP

Send a stream of bytes through a virtual “pipe”

Utilizes sequence numbers, acknowledgement, timeout, retransmission…

51


TCP/IP for Voice Over IP

Good for session management

H.323 / ASN.1 built on TCP/IP

Sub-optimal for near-realtime audio transport

Latency

Jitter

Aside: Utilized for Skype

52


UDP = User Datagram Protocol

Transport protocol:

One of the core protocols of the Internet protocol suite: 20% of all traffic

Does not provide the reliability and ordering guarantees of TCP

Datagrams may arrive out of order or be dropped by the network

Datagram transmission is stateless in the network

Lower overhead = faster, more efficient, suited to time-sensitive comms

UDP supports many application protocols including DNS and RTP

53


UDP for Voice Over IP

OK for session management

End-parties need reliable communication about session status

SIP utilises UDP with retry behaviour to withstand packet loss

UDP offers faster setup time than TCP/IP

Suited to near-realtime audio transport

Utilized by RTP

Better latency than TCP/IP (though not ideal)

Better jitter than TCP/IP (though not ideal)

Packet loss causes poorer audio transmission than TCP/IP

54

DRAFT V03DRAFT V03 Internet Protocols for Voice Over IPInternet Protocols for Voice Over IP

Internet vs. PSTN

Internet has smart terminals and “dumb” network

PSTN has “dumb” terminals and smart network

PSTN dedicates virtual connections for audio and session

Internet normally creates connections on an as-needed basis

Internet protocols emerging for traffic shaping suited to telephony

Network tools also emerging for banishing and punishing telephony

VoIP Protocols

56


VoIP Protocols

Overview of the landscape of VoIP protocols

Major IETF standards: SIP, RTP, RTCP

Major ITU-T standards: H.323 family

57

DRAFT V03DRAFT V03 VoIP Protocols: IETFVoIP Protocols: IETF

Protocol Description

SIP Session Initiation Protocol Session management on UDP

RTPSRTP

Real-time Transport ProtocolSecure RTP

Audio/video media delivery on UDP

RTCPSRTCP

Real-time Transport Control ProtocolSecure RTCP

Out-of-band control protocol for RTP

58

DRAFT V03DRAFT V03 VoIP ProtocolsVoIP Protocols

About SIP

IETF Protocol

Standard for initiating, modifying, and terminating a user session that may involves media elements such as voice, video, instant messaging etc.

Used widely in telephony environments

Supported by numerous IVR platforms

Accepted in 2000 as the signalling protocol of the IMS architecture

Other uses

MRCP v2: Media Resource Control Protocol for Speech Recognition and Text-to-Speech

Microsoft Messenger

59

DRAFT V03DRAFT V03 VoIP Protocols: ITU-TVoIP Protocols: ITU-T

Protocol Description

H.323Umbrella recommendation for audio-visual comms on any packet networkReferences the following specifications

H.225.0Protocol to describe call signaling, the media (audio and video), the stream packetization, media stream synchronization and control message formats

H.245

Control protocol for multimedia communication with messages and procedures used for opening and closing logical channels for audio, video and data, capability exchange, control and indications

H.235 Describes security in H.323

H.329Describes dual stream use in videoconferencing, usually one for live video, the other for presentation

60

DRAFT V03DRAFT V03 VoIP ProtocolsVoIP Protocols

About H.323

Based on ISDN Q.931

Suited to internetworking between IP and ISDN / QSIG

Similar call model to ISDN

Used widely in telephony environments

Telecommunications backbones

Other uses

Microsoft NetMeeting

61

DRAFT V03DRAFT V03 Today’s FocusToday’s Focus

We’ll focus on SIP today

SIP: Session Initiation Protocol

63


Session Initiation Protocol

Components of the SIP architecture

SIP Messaging

Standard SIP exchanges

Ring, hold, answer, transfer, consultative transfer, conferencing

SIP addresses

SIP working with RTP for Voice

64

DRAFT V03DRAFT V03 SIP OverviewSIP Overview

Session Initiation Protocol (SIP)

There are many applications of the Internet that require the creation and management of a session, where a session is considered an exchange of data between an association of participants. The implementation of these applications is complicated by the practices of participants: users may move between endpoints, they may be addressable by multiple names, and they may communicate in several different media - sometimes simultaneously. Numerous protocols have been authored that carry various forms of real-time multimedia session data such as voice, video, or text messages. The Session Initiation Protocol (SIP) works in concert with these protocols by enabling Internet endpoints (called user agents) to discover one another and to agree on a characterization of a session they would like to share. For locating prospective session participants, and for other functions, SIP enables the creation of an infrastructure of network hosts (called proxy servers) to which user agents can send registrations, invitations to sessions, and other requests. SIP is an agile, general-purpose tool for creating, modifying, and terminating sessions that works independently of underlying transport protocols and without dependency on the type of session that is being established.

65



http://www.ietf.org/rfc/rfc3261.txt

Protocol developed by the IETF MMUSIC Working Group (now SIP)

Scope: initiate, modify and terminate an interactive user session that involves voice and multimedia elements such as video, instant messaging and games.

SIP 2.0 published as RFC 3261 in 2002 Initial release of SIP 1.0 as RFC 2543 in 1996 (now obsolete)

SIP enables device-to-device communication with media communication via other protocols SDP: Session Description Protocol - RFC 2327 (describe media capabilities)

RTP: Real-time Transport Protocol - RFC 3550 (transport audio, video, media)

RTCP: Real-time Transport Control Protocol - RFC 3550 (control transport of media)

Standard protocols with high level of product interoperability

http://www.ietf.org/rfc/rfc3261.txt

66



SIP is an “application-layer” control protocol in Internet stack

SIP is an device-to-device, client-server session signalling protocol

SIP establishes sessions for voice and other media

Allows integration with others services: web, email, IM…

Allows presence and mobility services

67


Applications of SIP

SIP can convey arbitrary payload

Session description

Instant messages

Pictures (e.g. picture of the caller)

Speech recognition control

Web pages

68

DRAFT V03DRAFT V03 SIP Overview: DevicesSIP Overview: Devices

CiscoCisco

xTenxTenSiemensSiemens

AvayaAvaya

BlackBerryBlackBerry

Express TalkExpress Talk

69


Network Devices

SIP Proxy Server

Intermediary to relay call signalling

SIP Redirect Server

Redirects callers to other servers

SIP Registrar

Accept registration requests from users

Maintains user’s whereabouts

SIP IVR

SIP PBX

70

DRAFT V03DRAFT V03 SIP CommunicationsSIP Communications

SIP Jargon

User Agent Client = Initiates a communication

User Agent Server = Respondent to a communication

Note: device can be both client and server in a single session

Examples:

Desktop phone is both a client (makes calls) and server (receives calls)

71


SIP Addresses

SIP address can make you globally reachable Callees bind to this address using SIP REGISTER method

Callers use this address to establish real-time communication with callees

SIP address is a URI address format: sip:[email protected]

sip:[email protected]

sip:[email protected]?subject=wassup

Can embed in web pages or place on your business card

Highlighted text is the public identifier

SIP URI address contents Must include host

May include user name

May include the port number

May include others parameters (e.g., transport)

72


Protocol design

Similar protocol to HTTP and SMTP

Transmission via UDP messages

Human-readable messages

Simple interaction mechanism

User Agent “A” sends a Control Message to User Agent “B”

User Agent “B” sends a Response Code to User Agent “A”

Retry in the event of communication failure

73


SIP Methods (Control Messages)

INVITE = invite a user agent to a session

ACK = acknowledge a communication

OPTIONS = query servers about their capabilities

REGISTER = register with a SIP Registrar

BYE = terminate a session

CANCEL = cancel a session

74


SIP Response Codes

1xx: Provisional -- request received, continuing to process the request

2xx: Success -- the action was successfully received, understood, and accepted

200 OK

3xx: Redirection -- further action required to complete the request

4xx: Client Error -- the request contains bad syntax or cannot be fulfilled

5xx: Server Error -- the server failed to fulfil an apparently valid request

6xx: Global Failure -- the request cannot be fulfilled at any server

75


SIP Extensions (selection amongst many)

INFO = carry session-related control information

RFC 2976

e.g. ISUP and ISDN signalling messages

REFER = refer the recipient to a new resource

RFC 3515

e.g. Call transfer

76

DRAFT V03DRAFT V03 SIP: Simple Peer-to-Peer SessionSIP: Simple Peer-to-Peer Session

“A” “B”

Simple peer-to-peer SIP session

Assumptions

“A” knows address of “B”

“A” and “B” can see each other on the network

Audio communication

Humans are using “A” and “B”

77


“A” “B”

INVITE From:A To:B A-SDP

100 TRYING

180 RINGING Play ring-tone to user BPlay ring-tone to user A

User makes a call to “B”

B User answers call200 OK B-SDP

RTP Audio Stream

RTP Audio Stream

ACK

A hears B B hears A

B hangs upBYE

200 OK B terminates RTP audio

- - - Call Established - - -

A terminates audio

- - - Session Over - - -

- - - Waiting for answer - - -

78


“A” “B”


100 TRYING

180 RINGING Play ring-tone to user BPlay ring-tone to user A



RTP Audio Stream

RTP Audio Stream

ACK

A hears B B hears A

B hangs upBYE



A terminates audio



SIP MessagesStatus codes

RTP

79


SIP Message: Example

INVITE sip:[email protected]:5077 SIP/2.0Via: SIP/2.0/UDP 10.0.0.3:5060From: <sip:[email protected]>;tag=40A0C340-2BCTo: <sip:[email protected]>Date: Fri, 07 Jul 2006 01:58:55 GMTCall-ID: [email protected]: timer,100relMin-SE: 1800Cisco-Guid: 4217155242-210899419-2988540394-3730825786User-Agent: Cisco-SIPGateway/IOS-12.xAllow: INVITE, OPTIONS, BYE, CANCEL, ACK, PRACK, COMET, REFER, SUBSCRIBE, NOTIFY, INFOCSeq: 101 INVITEMax-Forwards: 6Remote-Party-ID: <sip:[email protected]>;party=calling;screen=yes;privacy=offTimestamp: 1152237535Contact: <sip:[email protected]:5060>Expires: 180Allow-Events: telephone-eventContent-Type: application/sdpContent-Length: 235

<<SDP header goes here>>

Header

To & From

Unique call ID

User agent details

Transmission info

Multi-part media content

SDP removed

80


SDP – Session Description Protocol

RFC 2327

Describe media capability of a SIP user agent

v=0o=Holly-HVG-4-2 2890844526 2890842809 IN IP4 10.0.0.113s=sip call from the hvgc=IN IP4 10.0.0.113t=0 0m=audio 11946 RTP/AVP 8 101c=IN IP4 10.0.0.113a=rtpmap:0 PCMU/8000a=rtpmap:8 PCMA/8000a=rtpmap:101 telephone-event/8000a=fmtp:101 0-16

v=0o=Holly-HVG-4-2 2890844526 2890842809 IN IP4 10.0.0.113s=sip call from the hvgc=IN IP4 10.0.0.113t=0 0m=audio 11946 RTP/AVP 8 101c=IN IP4 10.0.0.113a=rtpmap:0 PCMU/8000a=rtpmap:8 PCMA/8000a=rtpmap:101 telephone-event/8000a=fmtp:101 0-16

Agent Description

Audio Capability #1RTP Protocol

Agent Address

G.711 u-law & A-law8000Hz

DTMF events 0123456789ABCD*#

81


SDP Protocol Structure

v= (protocol version)o= (owner/creator and session identifier).s= (session name)i=* (session information)u=* (URI of description)e=* (email address)p=* (phone number)c=* (connection information)b=* (bandwidth information)

One or more time descriptionsz=* (time zone adjustments)k=* (encryption key)a=* (zero or more session attribute lines)

Zero or more media descriptionsm= (media name and transport address)i=* (media title)c=* (connection information - optional if

included at session-level)b=* (bandwidth information)k=* (encryption key)a=* (zero or more media attribute lines)

Time descriptiont= (time the session is active)r=* (zero or more repeat times)

* = optional

82

DRAFT V03DRAFT V03 SIP RegistrationSIP Registration

“A”“A”

Registration Process

SIPRegistrar

SIPRegistrar

REGISTER

200 OK

- - - Repeat regularly (5, 60+ min) - - -

REGISTERVia: IP:HostFrom: “Andrew Hunt” <sip:[email protected]>To: <sip:asterisk.holly-connects.com:5060>CallID: <<something unique>>Expires: 3600

REGISTERVia: IP:HostFrom: “Andrew Hunt” <sip:[email protected]>To: <sip:asterisk.holly-connects.com:5060>CallID: <<something unique>>Expires: 3600

83

DRAFT V03DRAFT V03 SIP: SIP Session Via a ProxySIP: SIP Session Via a Proxy


100 TRYING

180 RINGING Play ring-tone to user B

Play ring-tone to user A



RTP Audio Stream

RTP Audio Stream

ACK

A hears B

B hears A

BYE



A hangs up



“A”“A” SIP ProxySIP Proxy “B”“B”


100 TRYING

180 RINGING

200 OK B-SDP

ACK

BYE

200 OK

A terminates RTP audio

84

DRAFT V03DRAFT V03 SIP: SIP Proxy ChainingSIP: SIP Proxy Chaining

“A”“A”

“B”“B”

SIP ProxySIP Proxy

SIP ProxySIP Proxy

SIP ProxySIP Proxy

85

DRAFT V03DRAFT V03 SIP ProxySIP Proxy

SIP Proxy Function

Serve as rendezvous point at which callees are reachable

Perform routing function

Select the next hop or hops when chaining

Forking: try multiple destinations in parallel or sequence

Avoid loops when chaining

Available capabilities

Programmable routing decisions & tables

Least-cost routing

Firewall traversal

Direct certain calls to PSTN via gateway (e.g. 911, local calls)

86

DRAFT V03DRAFT V03 SIP: Session Via a PBXSIP: Session Via a PBX


100 TRYING

180 RINGING Play ring-tone to user B

Play ring-tone to user A



RTP Audio Stream

RTP Audio StreamACK

A hears B

BYE

200 OK

B terminates RTP audio




“A”“A” SIP PBX(e.g. Asterisk)

SIP PBX(e.g. Asterisk) “B”“B”

INVITE From:P To:B P-SDP

100 TRYING

180 RINGING

200 OK B-SDP

ACK

BYE

200 OKA terminates RTP audio

REGISTER

OK REGISTER

OK

RTP Audio Stream

RTP Audio Stream

B hangs up

87

DRAFT V03DRAFT V03 SIP: Session Via a PBX with Redirect (Direct Audio Link)SIP: Session Via a PBX with Redirect (Direct Audio Link)

BYE

200 OK

B terminates RTP audio



“A”“A” SIP PBX(e.g. Asterisk)

SIP PBX(e.g. Asterisk) “B”“B”

BYE

200 OKA terminates RTP audio

B hangs up

(Re)INVITE From:P To:B A-SDP

200 OK

RTP Audio Stream

(Re)INVITE From:P To:A B-SDP

RTP Audio Stream

- - - Call Continues - - -

88

DRAFT V03DRAFT V03 SIP-TDM GatewaySIP-TDM Gateway

Translate signalling messages

To/From traditional telephony and VoIP

e.g. ISUP SIP/RTP

Support heterogeneous environments

Staged migration to VoIP

Numerous gateway products available

PSTN IP

TDM

/ISDN

SIP/R

TP

VoIPGateway

89


Signalling gatewaySignalling gateway

Media Gateway Controller

Media Gateway Controller

Media GatewayMedia Gateway

PSTN IP

RTP

SIP

TDM

ISDN

90


Terminates RTP audio


TDM-VoIPBridge

TDM-VoIPBridgeSIP UASIP UA

Q.931 RELEASE COMPLETE

Terminates RTP audio

Pick-up

INVITE 555 1234

100 TRYING

RTP Audio Stream

RTP Audio Stream

180 RINGING

Q.931 SETUP

Q.931 CALL PROCEEDING

Play ring-tone

User makes a call

Q.931 PROGRESS

Q.931 CONNECT

Phone rings

200 OK

ACK

BYEQ.931 DISCONNECT

200 OK

Hang up

Q.931 RELEASE

- - - Call in Progress - - -

91

DRAFT V03DRAFT V03 SIP: Sending Auxiliary InformationSIP: Sending Auxiliary Information

User makes a call

“A”“A”

INVITE sip:[email protected]:5077 SIP/2.0<<SIP header goes here>>Content-type: multipart/mixed; boundary="gc0p4Jq0M2Yt08jU534c0p"MIME-version: 1.0

This is a multi-part message in MIME format.--gc0p4Jq0M2Yt08jU534c0pContent-Type: application/sdpContent-Length: 235

<<SDP header goes here>>--gc0p4Jq0M2Yt08jU534c0pContent-type: image/jpeg PGh0bWw+CiAgPGhlYWQ+CiAgPC9oZWFkPgogIDxib2R5PgogICAgPHA+VGhpcyBpcyB0aGUgYm9keSBvZiB0aGUgbWVzc2FnZS48L3A+CiAgPC9ib2R5Pgo8L2h0bWw+Cg==--gc0p4Jq0M2Yt08jU534c0p

“Brad calling”

92

DRAFT V03DRAFT V03 SIP: Failed EstablishmentSIP: Failed Establishment

“A” “B”

INVITE From:A To:B A-SDPUser makes a call to “B”

INVITE From:A To:B A-SDPTimeout

- - - Retries - - -

“B” gives up

INVITE From:A To:B A-SDPTimeout

Timeout

RTP: Real-time Transport Protocol

94


RTP: Real-time Transport Protocol

Really real-time?

RTP session overview

RTP packets

RTP and network transmission

95

DRAFT V03DRAFT V03 Real-time Transport ProtocolReal-time Transport Protocol

Standard for real-time transport over IP networks

Streaming audio and video

Utilised in SIP/RTP and H.323

Adopted by 3GPP for next generation cellular telephony

Widespread use in streaming: QuickTime, Real, Microsoft

RTP assumes

Network is dumb and imperfect, end-points are smart

Network may exhibit delays, jitter, packet loss etc.

Real-time Transport Protocol is NOT REAL-TIME

No end-to-end protocol, including RTP, can ensure in-time delivery. This would require the support of lower layers (switches, routers etc.)

RTP provides functionality suited for carrying real-time content, e.g., a timestamp and control mechanisms for synchronizing different streams with timing properties

96


One RTP session transmits one media type

Audio / voice

Video

Multi-media requires multiple RTP sessions

RTP session:

Implements a particular RTP profile

Includes an RTP data flow Transports a single media type according to one or more payload formats e.g. audio in G.711 format

Includes an RTP control protocol flow

Providing reception quality feedback, user information, etc.

Associates: Source and destination IP addresses A pair of UDP ports: one for RTP, one for RTCP

97


Media Content

Supports mixinge.g. for audio conferencing

Detect packet lossTime of last payload sample

98


Source

Destination

Playback

Buffer

Network

Packet loss

Out-of-order recovery

Late packet loss

Network Issues and Design

100


Network Issues and Design

Quality of Service

Packet loss, Latency, Jitter

Managing Quality of Service

Network quality

MPLS

Silence suppression

Local vs. long distance

Network design

Firewalls: NAT & STUN

101

DRAFT V03DRAFT V03 Quality of ServiceQuality of Service

QoS Definition

Probability of the network meeting a given traffic contract

Informally refers to the probability of a packet succeeding in passing between two points in the network within its desired latency period

102


What can go wrong as packets go from “A” to “B”?

Dropped packets: routers might fail to deliver (drop) some packets if they arrive when their buffers are full. Some, none, or all of the packets might be dropped, depending on the state of the network, and it is impossible to determine what happened in advance.

Delay: it may take a long time for a packet to get from “A” to “B” because it gets held up in long queues or takes a less direct route to avoid congestion

Jitter: packets from source will reach the destination with different delays, sometimes by taking different routes

Out-of-order delivery: different packets may take different routes with enough difference in delay to change the order of arrival

Error: sometimes packets are misdirected, or combined together, or corrupted, while en route

103


104


QoS issues can have a major impact on real-time media streaming

Packet loss

Missing packet replaced by silence

Jitter & out-of-order

Abrupt and unwanted variation in packet arrival timing

Arrival of packets out-of-order

Original

Packet loss

Jitter

105


Measure QoS on routers and end-points

QoS tends to degrade with network size and congestion

Managing Quality of Service

Generously over-provision a network – expensive and does not scale

Reserve network resources: e.g. RSVP = Resource Reservation Protocol

DiffServ: Differentiated services for bulk flows (e.g. packets from a university)

Multi-Protocol Label Switching (MPLS): emulates some properties of a circuit-switched network over a packet-switched network.

Traffic shaping: control computer network traffic to optimize or guarantee performance, low latency, and/or bandwidth. Traffic shaping deals with concepts of classification, queue disciplines, enforcing policies, congestion management, quality of service (QoS), and fairness.

Silence suppression

106


QoS on Local Network

QoS generally not a major issue in single-site deployments

Dedicate separate switches for voice and data traffic

Provide redundant networks for voice and data traffic

Gigabit Ethernet

Monitor QoS

Alarm network outages

107

DRAFT V03DRAFT V03 FirewallsFirewalls

Network Address Translation (NAT): re-writing the source and/or destination addresses of IP packets as they pass through a router or firewall

Aka network masquerading or IP-masquerading

Firewalls use NAT to enable multiple hosts on a private network to access the Internet using a single public IP address

Common in home and SOHO routers

SDP addresses are not translated by NAT

STUN (Simple Traversal of UDP over NATs, RFC 3489):

Network protocol allowing clients behind NAT to find its (a) public address, (b) the type of NAT and (c) the internet side port associated by the NAT with a particular local port.

Info is used to set up UDP communication between two machines both behind NAT routers

VoIP and Speech Recognition

109


VoIP and Speech Recognition

MRCP v1 & v2

Impact of CODECs

Network issues: packet loss, latency

110

DRAFT V03DRAFT V03 MRCP: Media Resource Control ProtocolMRCP: Media Resource Control Protocol

MRCP v1

IETF protocol

Client control of speech resources

Speech recognition

Text-to-speech

MRCP structure is similar to HTTP and SIP

Request by client in header+body format

Response by server

Media delivery typically via RTP

Widely supported by VoiceXML Platforms

Leverages existing W3C standards for speech recognition and TTS markeup

111

DRAFT V03DRAFT V03 MRCP: Media Resource Control ProtocolMRCP: Media Resource Control Protocol

MRCP v2

IETF protocol to supersede MRCP v1

Broader client control of speech resources

Adds speaker verification and speaker identification

Adds recording

Utilizes SIP+SDP to establish the media pipe

112

DRAFT V03DRAFT V03 VoIP and Speech RecognitionVoIP and Speech Recognition

Speech Recognition and CODECs

Lossless CODECs do not affect speech recognition accuracy

No loss of information

Lossy CODECs can affect speech recognition accuracy

Greater compression tends to cause great degration

Speech recognizers are generally very reliable with widely deployed CODECs

Mobile telephony has extensive compression

Speech recognition trained explicitly for mobile performance

DSR Aurora: Distributed Speech Recognition

CODEC specialized for the requirements of speech recognition

Promoted for mobile carrier usage

113

DRAFT V03DRAFT V03

Speech Recognizer

VoIP and Speech RecognitionVoIP and Speech Recognition

Speech Recognition and Latency

Speech recognition not as sensitive to latency as humans

Late packet is better than no packet

Speech recognizers have extensive buffers for non-real-time processing

Note: excessive latency (>1sec) can cause caller perceived service issues

BufferRTP Result

114

DRAFT V03DRAFT V03

Speech Recognizer

VoIP and Speech RecognitionVoIP and Speech Recognition

Speech Recognition and Packet Loss

Speech recognition are sensitive to packet loss

ASR can use packet loss information to minimize error reduction

BufferRTP Result

VoIP and Mobile Telephony

116


VoIP and Mobile Telephony

Analog mobile

Digital mobile

3G and SIP

IMS – IP Multi-Media Subsystem Architecture

117

DRAFT V03DRAFT V03 Mobile TelephonyMobile Telephony

Analog Mobile

Experimental systems from 1920s

“1G” – 1st Generation

AMPS: Advanced Mobile Phone Services

Analog transmission

1978: Trial in Chicago

1979: Commercial launch in Japan

1981: Commercial launch in Sweden, Norway, Denmark, Finland

1983: Commercial launch in Chicago

Issues: limited capacity, fraud, subscriber volume…

118


2G – 2nd Generation Mobile

Objectives achieved

Digital technology

Increased capacity

Greater security against fraud

Global roaming

Advanced services

Lower power = smaller handsets = longer battery life

Many standards evolved (examples)

GSM: Pan-European standard that spread globally

CDMA: Americas and parts of Asia (aka PCS)

PDC: Japan

Limitations:

Optimized for voice – not suited to data

119


2.5G – Stepping Stone from 2G to 3G

2G system with both packet switching (for data) and circuit switching (for voice)

2.5G is a marketing term – not a standard

Objectives achieved

Re-use of much 2G infrastructure (GSM & CDMA)

Data rate of 144 kbit/sec or better

Used for sending photos and much more

120


3G – 3rd Generation Mobile

Combines high-speed mobile access with Internet Protocol (IP) based services

Covers range of network technologies

WCDMA, CDMA2000, UMTS, EDGE

Data rate: 384kbps for mobile systems and 2Mbps for stationary systems

Enables video, TV, images, music, games, location services…

121


3GPP – 3rd Generation Mobile

Collaboration agreement (Dec-98) between ETSI (Europe), ARIB/TTC (Japan), CCSA (China), ATIS (North America) and TTA (South Korea).

Goal: global 3G specification within the scope of the ITU's IMT-2000 project

3GPP specifications are based on evolved GSM specifications

Now generally known as the UMTS system

Introduced IMS…

122


IMS – IP Multi-Media Sub-System

Emerged in 3GPP Release 5 with following enhancements

Principles

Access independence: work with fixed, mobile or wireless networks

Different network architectures: implement on operator-selected architectures

Terminal and user mobility: provides terminal mobility (roaming)

Extensive IP-based services: offer just about any IP-based service. VoIP, push-to-talk over cellular (POC), multiparty gaming, video conferencing, messaging, community services, presence information, content sharing…

123


IMS is Built on SIP

3GPP Variant of SIP

Application servers for SIP session management

Caller ID, call waiting, call forwarding, transfer, call blocking, interception, announcements, conferencing, voice-mail, SMS…

CSCF: Call Session Control Function and other functions by SIP

Media Resource Function (MRF): SIP end-point with IVR-like functionality

TDM-VoIP gateways to bridge to fixed and mobile telephony

Who’s in control?

TDM = dumb terminals, smart network

Internet VoIP = smart terminals, dumb network

IMS = dumb/smart terminals, smart network

Closing

125

DRAFT V03DRAFT V03 Module ObjectivesModule Objectives

Go home!!

126

DRAFT V03DRAFT V03 Further InformationFurther Information

Web sites

IETF: http://www.ietf.org/html.charters/sip-charter.html

SIP Tutorial: http://www.iptel.org/sip/siptutorial.pdf

SIP Home Page: http://www.cs.columbia.edu/sip/

SIP Forum: http://www.sipforum.org/

Asterisk PBX: http://www.asterisk.org/

VoIP Wiki Reference: http://www.voip-info.org/wiki/

SIP Knowledge: http://www.sipknowledge.com/SIP_RFC.htm

SIP FAQ: http://www.sipknowledge.com/faq_main.htm

SIP Tech Portal: http://www.tech-invite.com/

http://www.ietf.org/html.charters/sip-charter.html

http://www.iptel.org/sip/siptutorial.pdf

http://www.cs.columbia.edu/sip/

http://www.sipforum.org/

http://www.asterisk.org/

http://www.voip-info.org/wiki/

http://www.sipknowledge.com/SIP_RFC.htm

http://www.sipknowledge.com/faq_main.htm

http://www.tech-invite.com/

Thank you!!!Thank you!!!

Documents

SpeechTEK 2006 – Voice Over IP Tutorial Andrew Hunt, Ph.D. VP Engineering, Holly Connects Andrew Hunt, Ph.D. VP Engineering, Holly Connects