Transcript Paper 1 - UCL Computer Science

Multimedia over Internet
 Paper 1
H. Schulzrinne, "A comprehensive multimedia control architecture for
the Internet", Proc. of the Int. Workshop on Network and Operating
System Support for Digital Audio and Video (NOSSDAV), (St. Louis,
Missouri), May1997.
 Paper 2
W. Jiang, J. Lennox, H. Schulzrinne and K. Singh, "Towards Junking
the PBX: Deploying IP Telephony", in Proc. International Workshop
on Network and Operating System Support for Digital Audio and
Video (NOSSDAV), (Port Jefferson, New York), Jun. 2001
Outline
 Overview of Internet
 Paper 1





Introduction
Session Initiation Protocol (SIP)
Real-Time Stream Protocol (RTSP)
Combining SIP and RTSP
Description of Multimedia Presentations
 Paper 2





Telephone Network
IP telephony
Architecture
PSTN Inter-operability
Other Issues
 Summary and Future Work
Overview of Internet
OSI
TCP/IP
Application
Presentation
Application
Session
Transport
Transport
Network
Internet
Data Link
Subnet
Physical
Overview of Internet
 The physical layer
This layer defines the type of physical signals ( electrical,
optical, etc.), as well as the type of media (wires, coaxial cable,
satellite, etc.).
 The data link layer
Common examples of data link control protocols are the HDLC,
SDLC, and PPP.
 The network layer ( Internet Protocol – IP)
 The transport layer ( TCP/UDP)
 The application layer
Telnet, ftp, SMTP, HTTP, NNTP, LDAP, Several multimedia
protocols ( SIP, RTP, H.323, etc. )
Overview of Internet
Ping
Telnet
FTP
H.323
SIP
RTSP
TCP
RSVP
S/MGCP
/NCS
RTP/
RTCP
UDP
OSPF
ARP
ICMP
IP
Link Layer
IGMP
RARP
Paper 1
A comprehensive multimedia control
architecture for the Internet
Henning Schelzrinne
[email protected]
+1 212 939 7042
Dept. of Computer Science
Columbia University
New York, NY 10027
Introduction
 In this paper, he present two independent, but
interacting protocols that initiate and control stored,
live and interactive multimedia sessions in the
Internet.
 The protocols support the following scenarios:
Phone call
Invitation to a multi-party conference
Near video-on-demand
Video-on-demand
Virtual presentations
Distributed digital editing
Combining stored, live and interactive multimedia
Introduction
 Session Initiation Protocol (SIP)


Inviting participants to a multimedia session
Establish and control multimedia conferences
 Real-Time Stream Protocol (RTSP)


Control playback and recording for stored continuous
media
Control delivery of stored and live streaming multimedia
content
Session Initiation Protocol
(SIP)
 Conference control applications use SIP to
invite humans and media servers into a
multicast conference or establish a two-party
phone call.
 The conference initiation phase has to
accomplish three goals:



locate the terminal (phone, workstation, mobile phone,
answering machine, … ) where the called party can be
reached.
agree on a set of media and possible encodings for
communication
determine if the called party wants to be reached.
Major Features of SIP
 User location
Determination of the end system to be used for communications.
 User capabilities
Determination of the media and media parameters to be used.
 User availability
Determination of the willingness of the called party to engage in
communications.
 Call setup
Establishment of call parameters at both called and calling party.
 Call handling
Including transfer and termination of calls.
Names and Addresses
 A name is an identification of an entity ( independent of its
physical location), such as a person, and applications
program, or even a computer.
 An address is also an identification but it reveals additional
information about the entity, principally information about
its physical or logical placement in a network.
 The IP Address ( 32 bits )
Class A
Class B
Class C
Multicast format
0 Network (7 bits)
10
110
1110
Local address ( 24 bits)
Network address (14 bits)
Local address (16 bits)
Network address (21 bits)
Multicast address (28 bits)
Local (8 bits)
Session Initiation Protocol
(SIP)
 SIP chose an email-like identifier of the form
user@domain or user@IP_address.
 The domain name can be either the name of
the host that a user is logged in at the time, an
email address or the name of a domainspecific translation service.
SIP address resolution
Address is
SIP server?
N
Address is
SMTP server?
Y
N
Y
Y
failure
Get address
(VRFY,EXPN)
forward?
N
Y
same as
before ?
accept ?
Y
N
Busy, no answer
reject
success
Send MIME
message
N
SIP Redirect Server
cse.psu.edu
Redirect Server
Location Server
alice
INVITE [email protected]
bob@play
200 OK
play
SIP Proxy Server
cse.psu.edu
Proxy Server
Location Server
bob
alice
play.cse.psu.edu
play
bob@play
SIP Forking Proxy
cse.psu.edu
Location Server
Proxy Server
alice
bob
run.cse.psu.edu
jump.cse.psu.edu
run
jump
bob@jump
Other issues
 Choosing Terminals
Many people have several ways of being reached, including a telephone,
email, fax, or a pager. A SIP server can return a descriptive list of
alternative terminals, their capabilities and addresses.
 Locating Callees
In a local area, a person may move around from terminal to terminal. A
SIP can work over a connectionless transport protocol and multicast a
“search” for a particular party.
 Negotiating Media Types and Encodings
The SIP INVITE request to join a conference or phone call contains a
listing of the media types and associated encodings that the calling party
is willing to use. The called party simply responds with a subset of
media types and encodings that it is willing to use.
Real-Time Stream Protocol
(RTSP)
 RTSP initiates and controls delivery of stored and live
multimedia content to both unicast and multicast
destinations.
 The Real-Time Protocol (RTP) is designed to support real
time traffic, which provides services that include payload
type identification, sequence numbering, time-stamping, and
delivery monitoring.
 The Real-Time Control Protocol (RTCP) is a control
component. Both data sender and receivers periodically
multicast RTCP messages to monitor network quality.
Real-Time Protocol (RTP)
512 kbit/s
384 kbit/s
Transit
Network
translator
384 kbit/s
Real-Time Protocol (RTP)
64 kbit/s each
64 kbit/s
Transit
Network
( Combine 192 kbit/s
to 64 kbit/s )
mixer
64 kbit/s
The basic operation of RTSP
HTTP GET
session description
web
server
SETUP
PLAY
client
RTP audio
RTP video
RTCP
PAUSE
TEARDOWN
media
server
The basic operation of RTSP
 First, the client should obtain a description of the
multimedia presentation. The description can be
retrieved by HTTP or ftp. It can be different format.
 Then the client will initiates a session with the SETUP
request to the media server. The SETUP request also
indicates where the server is to send the data, if not
provided in the presentation description.
 The presentation itself can be controlled with PLAY,
RECORD and PAUSE.
 The client closed the session with the TEARDOWN
request.
Description of Multimedia Presentations
 A data structure to describe the session or
presentation they are initiating and controlling.
 SDF describes presentations as a hierarchy of
sequential, alternative and time-parallel streams.
 The design of SDF found its way into a proposed
SGML-based description called RTSL.
 RTSL is intentionally purely descriptive and
contains no scripting functionality.
Sample RTSP session description
<title>Twister</title>
<session>
<group language=en lipsync>
<switch>
<track type=audio
e=“PCMU/8000/1”
src=“rtsp://audio.example.com/twister/audio.en/lofi”>
<track type=audio
e=“DVI4/16000/2” pt=“90 DVI4/8000/1”
src=“rtsp://audio.example.com/twister/audio.en/hifi”>
</switch>
<track type=“video/jpeg”
src=“rtspu://video.example.com/twister/video”>
</group>
</session>
Combining SIP and RTSP
Internet conferencing example
Combining SIP and RTSP
Combining SIP and RTSP
Possible client conferencing architecture using SIP and RTSP
Paper 2
Towards Junking the PBX: Deploying
IP Telephony
Wenyu Jiang, Jonathan Lennox, Henning Schelzrinne and Kundan Singh
{wenyu,lennox, hgs,kns10}@cs.columbia.edu
Dept. of Computer Science
Columbia University
New York, NY 10027
Telephon e Network
 The goals of a telephone system



There had to be sufficient direct current flow to operate the customer’s
station sets.
Support dc/low-frequency call process signaling (dialing, ringing) and to
keep the signaling simple at the customer’s terminal.
Limit signal loss to acceptable levels such that the voice conversation
between the customers would appear as “natural” as possible.
 The telephone dialing plan
NXX-XXXX
N: 2~9
X: 0~9
Example of a call
Originating
Office
Local Station
Terminating
Office
Idle
Idle
Local Station
Idle
Line connect
Dial tone
Dial pulsing
Trunk connect
Start dial
Dial pulsing
Ringing
Ringback
Answer
Answer
Answer
Busy
IP Telephony
 Internet telephony is defined as the transport of
telephone calls over the Internet.
 Internet telephony integrates a variety of services
provided by the current Internet and the Public
Switched Telephone Network (PSTN)
infrastructure.
 Internet telephony employs a variety of protocols,
including RTP, H.323, MGCP, Megaco, SIP, etc.
IP Telephony
 Services and prices vary, with these offerings:






PC-to-PC calls
PC-to-telephone calls
Telephone-to-telephone calls
Fax service
E-mail
Voice messaging
 Examples of Service Providers:





Deltathree (www.deltathree.com )
Net2Phone (www.Net2Phone.com)
MediaRing (www.mediaring.com)
Dialpad (www.dialpad.com)
PhoneFree (www.phonefree.com)
Architecture
Architecture
 SIP server
 SQL database
 PSTN gateway
 User agents
 Media server
SIP proxy, redirect and registration server
storing the current network addresses and phone
numbers where the user can be reached.
connect the PBX to the LAN with a T1 trunk
allow users to interact with the system over IP
storage and delivery of announcements and voice
mail messages
 Unified messaging
 Conference server
 SIP-H.323 translator
centralized answering machine and
voice mail system
centralized audio/video conference server
a signaling gateway between SIP and H.323
User Database
 Every user of the system is given a unique identifier
of the form user@domain, also called a canonical
user identifier.
 The user information is stored in the SQL database
as the Primary User Table and indexed by the user
identifier.
 There are other tables in the MySQL database:


contact table
alias table
Incoming Calls
Incoming Calls
 Transform the callee address to a canonical user
identifier for database look up


host portion: erlang.cs.columbia.edu—cs.culumbia.edu
User name portion: aliasname mappingdial plan
 Retrieves user, contact, and policy information



Proxied or redirected
Authentication
Forking proxy
PSTN Inter-Operation
 Dialplan



On the gateway, define a voice over IP call-leg specifier(called a dial peer)
Direct-inward-dialing (DID) mode
No-DID mode
Mode
usage
advantages
DID
Dial directly
Simpler dialing from PSTN
No-DID
Dial extension
Supports more users
 Connecting to the PBX
PSTN Inter-Operation
 Security Issues

User registrations
user registrations need to be authenticated to prevent unauthorized
users from redirecting calls to themselves or elsewhere.

Remote callers
a local user may choose to force remote callers to be authenticated.
Our authentication goal is to establish a consistent mapping between
a caller’s SIP identity and her email identity. This ensures that the
SIP caller is indeed identical to the corresponding email address.

Access to the PSTN
we need to restrict access to the PSTN gateway to prevent “free”
calls.
Other Services
 Programmable Call Handling


The XML based Call Processing Language (CPL)
The SIP Common Gateway Interface
 Unified Messaging


Centralized voice mail
RTSP for storage and retrieval of voice messages
 Multi-Party Conferencing




A SIP conference server with audio and video capabilities
The canonical user identifier
The dynamic modification of the dialplan
The SQL database stores various conference attributes
Scalability
 Multiple conference servers can be installed, with
each running only tens of active conferences.
 For scaling proxy servers, make use of the DNS
SRV capability in SIP.
Summary
 Paper 1 describe two protocols that support the multimedia
conference, namely the SIP to establish and control
multimedia conferences and the RTSP to control delivery of
stored and live streaming multimedia content.
 Paper 2 describes the architecture of the Internet telephony
installation:






SIP server
SIP-PSTN gateway
RTSP media server
unified messaging server
conferencing server
SIP-H.323 translator.
Future Work
 The Internet multimedia architecture is still missing two
pieces, namely a floor control protocol and a shared
drawing protocol.
 “embedded application” that work behind the scenes of
web pages, games and virtual reality.
 Continue with integration of additional services.
 Build highly scalable systems.
 A commercial deployment involves many other issues
related to security, billing and quality of service.