Demystifying Multimedia Conferencing Over the Internet Using ...

Demystifying Multimedia Conferencing Over the Internet
Using the H.323 Set of Standards
James Toga, Emerging Products Division, Intel Architecture Labs, Intel Corporation
Hani ElGebaly, Emerging Products Division, Intel Architecture Labs, Intel Corporation

Index words: H.323, conferencing, Internet, multimedia, gatekeeper

new H.323 features such as supplementary services and
Abstract security facilitate this purpose.
The Telecommunication Sector of the International
Telecommunication Union (ITU-T) has developed a set of Introduction
standards for multimedia conferencing over packet-based
Recommendation H.323 [8] describes the procedures for
networks. These standards are aggregated under a
point-to-point and multipoint audio and video
standard umbrella termed Recommendation H.323.
conferencing over packet-switched networks. In addition
Recommendation H.323 describes terminals, equipment,
to video conferencing terminals, Specification H.323
and services for multimedia communication over networks
describes other H.323 entities including gateways,
such as the Internet.
gatekeepers, and MCUs. Gateways allow interoperation of
H.323 terminals and equipment carry real-time voice, H.323 systems with other audio/video conferencing
video, and data, in any combination thereof. Terminals systems on integrated services digital networks (ISDN),
signal calls using Q.931-derived procedures defined in plain old telephone systems (POTS), asynchronous
Recommendation H.225.0. After the call signaling phase, transfer mode (ATM), and other transports. Gatekeepers
terminals proceed to the call control phase where they provide admission control and address translation to
exchange capabilities and logical channel information H.323 endpoints. MCUs can engage more than two H.323
using the H.245 protocol defined in Recommendation endpoints in a centralized multipoint conference.
H.245. Once the call has been established, audio and
Recommendation H.323 comprises a number of related
video (if supported) are initiated. Both media types use
documents that describe terminals, equipment, services,
the Real Time Protocol (RTP) defined by the Internet
and interactions. Examples of other core standards that are
Engineering Task Force (IETF) as their Transport
referred to in Recommendation H.323 are H.225.0 [5]
Protocol. Procedures for audio and video packet format
(procedures for call signaling, media packet format, and
and transport are described in Recommendation H.225.0.
synchronization); H.245 [7] (procedures for capability
Recommendation H.323 allows for the use of a variety of
exchange, channel negotiation, and flow control); H.450.x
video codecs (e.g., H.261, H.263, H.263+) and audio
[11] (procedures for supplementary services); H.246 [8]
codecs (e.g., G.711, G.723.1). Data collaboration is also
(procedures for terminals’ interoperability through
allowed using Protocol T.120. We provide an overview of
gateways); and H.235 (security and encryption
these protocols and explain the H.323 call scenario.
procedures). Other referenced standards include media
Other entities such as gatekeepers, Multipoint Control codecs for audio (such as, G.711, G.723.1, G.729, and
Units (MCUs), and gateways are also addressed in G.722) and video (such as, H.261 and H.263).
Recommendation H.323. These entities allow network
The purpose of this paper is to present an overview of the
management, centralized multipoint, and interoperability
H.323 core components and functionality and the current
with other conferencing standards. We explain each of
industry trends with respect to H.323, such as IP
these entities briefly and provide some scenarios of their
telephony and corporate conferencing. Towards this goal,
interaction with the H.323 terminals.
we briefly describe the H.323 architecture, the
IP telephony has become an important driver for packet- responsibilities of H.323 entities, and basic call scenarios.
based communications. We address the role of H.323 The main obstacles to the adoption and deployment of the
procedures in deploying IP telephony, and describe how H.323 standard in industry are also explained. Finally,

Demystifying Multimedia Conferencing Over the Internet using the H.323. Set of Standards 1

Intel Technology Journal Q2’98

H.323 zone management, multipoint operation, telephony codecs, resolution, bitrate, and algorithm options are
services features, and security are highlighted. exchanged between terminals, using the H.245 protocol.
Terminals can open channels only with parameters and
H.323 Architecture options chosen from the intersection of the capability sets.
In general, the receiver always specifies what the
transmitter may send.
Scope of Recommendation H.323
The audio codec (G.711, etc.) encodes the audio signal
Video from the microphone for transmission and decodes the
Video I/O
H.261
received audio code that is output to the loudspeaker.
H.263 G.711 0 is the mandatory codec for an H.323 terminal. A
Audio terminal may be capable of optionally encoding and
G.711 Recv decoding speech using Recommendations G.722, G.728,
Audio I/O G.728 Path G.729, MPEG1 audio, and G.723.1. Since G.711 is a
G.729 Pkt
Delay H.225 high-bitrate codec (64Kb/s or 56Kb/s), it cannot be
G.723.1 Network
G.722
Layer carried over low-bitrate (< 56 kbps) links. G.723.1 [2] is
Interface
the preferred codec in this situation because of its
User App reasonably low rate (5.3Kb/s and 6.4 Kb/s). H.323
T.120, etc. terminals open logical channels using a common
capability that is supported by all entities and exchanged
during the H.245 capability exchange phase.
H.245
System The Data Channel supports telematic applications such as
Control electronic whiteboards, still image transfer, file exchange,
Q.931
User database access, audiographics conferencing, etc. The
Interface standardized data application for real-time audiographics
RAS conferencing is T.120. Other applications and protocols
may also be used via H.245 negotiation such as chatting
and fax.
Figure 1: H.323 architecture The System Control Unit (H.245, H.225.0) provides
The architecture of an H.323 terminal is shown in Figure signaling and flow control for proper operation of the
1. The scope of Recommendation H.323 is limited to the H.323 terminal. H.245 [7] is the media control protocol
definition of the media compression standard, packet that allows capability exchange, channel negotiation,
format, signaling, and flow control. Media capturing such switching of media modes, and other miscellaneous
as video capturing schemes, audio recording, or user data commands and indications. The H.225 standard describes
applications are outside the scope of Recommendation the Call-Signaling Protocol used for admission control
H.323. Initial H.323 implementations targeted IP and for establishing connection between two or more
networks; however, Recommendation H.323 supports terminals. The Connection Establishment Protocol is
alternative transports such as IPX. The Transport layer derived from the Q.931 specification [12]. The H.225.0
carries control and media packets over the network; there [5] Layer also formats the transmitted video, audio, data,
are no specific requirements for the underlying transport and control streams into messages for output to the
except for support for reliable and unreliable packet network interface, and it retrieves the received video,
modes. An example of a well-known transport is User audio, data, and control streams from messages that have
Datagram Protocol (UDP) over Internet Protocol (IP). been input from the network interface, using the Real
Time Transport Protocol (RTP) and its companion, the
The video codec (e.g., H.261, etc.) encodes the video Real Time Control Protocol (RTCP). The RTP performs
from the video source (i.e., camera) for transmission and logical framing, sequence numbering, timestamping,
decodes the received video code that is output to a video payload distinction, source identification, and
display. The mandatory video codec for an H.323 terminal occasionally, error detection and correction as appropriate
is H.261 [3] with quarter common intermediate format to each media type. The RTCP provides reporting and
(QCIF) resolution. (QCIF is a video picture size.) Other status that may be used by both senders and receivers to
codecs such as H.263 may be supported. H.263 [4] has correlate performance on the media streams.
better picture quality and more options. A terminal can
also support other picture sizes such as CIF and SQCIF.
During the terminal capability exchange phase, the video



H.323 Entities and Responsibilities of an H.323 endpoint to/from H.242 of an H.320
endpoint). This translation is specified in
Terminal 1 Terminal 2 Gatekeeper
Recommendation H.246. The gateway also performs call
setup and clearing on both the network side and the
Switched-Circuit Network (SCN) side. Translation
Packet Switched Network
between video, audio, and data formats may also be
performed in the gateway. In general, the purpose of the
gateway is to complete the call in both directions between
MC MP Terminal 3
Gateway the network endpoint and the SCN endpoint in a
MCU
transparent fashion.

Figure 2: H.323 entities The MCU is an endpoint on the network, which provides
the capability for three or more terminals or gateways to
Figure 2 shows the different entities that can be involved participate in a multipoint conference. It may also connect
in an H.323 conference. A gatekeeper is an optional two terminals in a point-to-point conference, which may
element in an H.323 conference that provides call control later develop into a multipoint conference. The MCU
services to the H.323 endpoints such as address consists of two parts: a mandatory Multipoint Controller
translation, admission control, bandwidth control, and (MC) and optional Multipoint Processors (MP). The MC
zone management. The gatekeeper may also provide other provides the capability for call control to negotiate with
optional functions such as call authorization and call all terminals to achieve common levels of communication.
accounting information. A gatekeeper cannot be specified It is this element that is required for all multipoint
as a destination in a call. Address translation is the method conferences. The MP allows mixing, switching, or other
by which an alias address (e.g., e-mail address) is processing of media streams under the control of the MC.
translated to a transport address. Admission control is a The MP may process a single media stream or multiple
way of limiting H.323 access to a network, or to a media streams depending on the type of conference
particular conference using Request, Admission, and supported. In the simplest case, an MCU may consist only
Status (RAS) messages defined in Recommendation of an MC with no MPs.
H.225.0. A gatekeeper also manages bandwidth allocation
to H.323 endpoints using RAS messages. Zone The following section provides an overview of the basic
management defines the scope of entities over which a operation of an H.323 endpoint in a point-to-point
gatekeeper has control. These include endpoints, conference without a gatekeeper and then with a
gateways, and MCUs. In addition, a gatekeeper can allow gatekeeper.
secure access to the conference using various
authentication mechanisms. Q.931 and H.245 messages H.323 General Operation
can be routed through the gatekeeper, and statistical Figure 3 shows call establishment and tear down steps
information about the calls in progress can be collected. between two H.323 endpoints without a gatekeeper. All of
Gatekeepers may also perform telephony service the mandatory Q.931 and H.245 messages exchanged are
operations such as call forwarding and call transfer. listed. Note that some of these messages may be
H.323 endpoints can interact with each other directly in a overlapped for increased performance. Each message has
point-to-point or multipoint conference if no gatekeeper is an assigned sequence number at the originating endpoint.
present. When a gatekeeper is present, all endpoints have Endpoint A starts by sending a Setup message (1) to
to register with it. endpoint B containing the destination address. Endpoint B
responds by sending a Q.931 Alerting message (2)
A gateway operates as an endpoint on the network that followed by a Connect message (3) if the call is accepted.
provides real-time, two-way communication between At this point, the call establishment signaling is complete,
H.323 terminals on the packet-based network and other and the H.245 negotiation phase is initiated. Both
ITU terminals on a switched-circuit network, or to another terminals will send their terminal capabilities (4) in the
H.323 gateway. Other ITU terminals include those terminalCapabilitySet message. The terminal capabilities
complying with Recommendations H.310 (B-ISDN), include media types, codec choices, and multiplex
H.320 (ISDN), H.321 (ATM), H.322 (GQoS-LAN), information. Each terminal will respond with a
H.324 (GSTN), H.324M (Mobile), and POTS. The terminalCapabilitySetAck (5) message. The terminals’
gateway provides the appropriate translation between capabilities may be resent at any time during the call. The
transmission formats (for example, H.225.0 of an H.323 Master/Slave determination procedure (6-8) is then
endpoint to/from H.221 of an H.320 endpoint) and started. The H.245 Master/Slave determination procedure
between communication procedures (for example, H.245 is used to resolve conflicts between two endpoints that can



both be the MC for a conference, or between two 1. GRQ 6. ACFARJ 11. DRQ
endpoints that are attempting to open bi-directional 2. GCFGRJ 7. ARQ 12. DCFDRJ
channels at the same time. In this procedure, two 3. RRQ 8. ACFARJ 13. URQ
endpoints exchange random numbers in the H.245 4. RCFRRJ 9. BRQ 14. UCFDRJ
masterSlaveDetermination message to determine the 5. ARQ 10. BCFBRJ
master and slave endpoints. H.323 endpoints are capable
of operating in both master and slave modes. Following Gatekeeper Cloud
master/slave determination procedures, both terminals
1 2
proceed to open logical channels (9-10). Both video and 2 1
audio channels are unidirectional while data is bi- 34 11 12 11 12 3 4
directional. Terminals may open as many channels as they 5 6 7 8
13 13 14
want. Only one logical channel is shown in Figure 3, yet 9 14 9 10
10
the same procedure applies if more channels are opened.
Q.931
1 1. Setup
2 2. Alerting Endpoint 1 Endpoint 2
3 H.245
4 3. Connect
5 4. termCapSet
A 4 B 5. termCapAck Figure 4: Gatekeeper interaction in an H.323 call
5 6. mSDet
6 Figure 4 shows messages exchanged between a gatekeeper
7. mSDetAck
7 and an H.323 endpoint. Before the conference starts, both
8.mSDetConfirm
8 endpoints look for a gatekeeper by multicasting a
9 9. openReq GatekeeperDiscovery (GRQ). Request. The gatekeeper
10 10. openAck will reply either with a GatekeeperConfirm (GCF)
9
11. endSession message or with a GatekeeperReject (GRJ) message. Both
10
11 12. ReleaseComplet endpoints will then register their alias names with the
11 gatekeeper using the RegistrationRequest (RRQ) message.
The gatekeeper will acknowledge by sending a
12 RegistrationConfirm (RCF) message or will deny it using
a RegistrationReject (RRJ) message. Registering alias
names with the Gatekeeper allows endpoints to call each
other using user-friendly addresses such as e-mail, etc.,
Q.931 messages
H.245 messages rather than the transport address. The discovery and
registration procedure is valid until the gatekeeper
Figure 3: Q.931 and H.245 messages exchanged between indicates otherwise.
two H.323 endpoints
An endpoint or gatekeeper can request the location of
Other H.245 control messages may be exchanged between another endpoint using its alias name by using a
the endpoints to change media format, request video key LocationRequest (LRQ) message, and the gatekeeper
frames, change the bitrate, etc. replies with a LocationConfirm (LCF) message containing
Termination of a call is initiated by one endpoint sending the resolved address for the alias name.
an endSession message (11). Endpoint B, on receiving the When a user places a call from an endpoint, the endpoint
endSession command, will respond with another starts by requesting admission from the gatekeeper using
endSession message (11) to Endpoint A. Endpoint A will an AdmissionRequest (ARQ) message. The gatekeeper can
finally send a Q.931 ReleaseComplete message (12), and accept (ACF) or deny the request (ARJ). If the call is
the call is terminated. accepted, the endpoint sends a Q.931 Setup message to
the remote destination. The recipient of the Setup message
in turn requests admission from its gatekeeper by sending
an ARQ. When the call is accepted, the Q.931 call
signaling sequence is completed followed by the H.245
message negotiation. The AdmissionRequest (ARQ)
message carries the initial bandwidth the endpoint requires
for the duration of the conference. If during H.245 logical
channel negotiation, an endpoint requires more



bandwidth, it issues a BandwidthRequest (BRQ) message conferencing between two or more users on the network.
to the gatekeeper. If the request is accepted, the In this application, the user is expecting the same services
gatekeeper replies with a BandwidthConfirm (BCF) as those provided by a telephone. The quality of service
message; otherwise, it replies with a BandwidthReject should be equal to or better than POTS. The Internet does
(BRJ) message. not appear to be readily addressing these issues; however,
there is work in progress at the corporate Intranet level to
When the call is terminated, both endpoints send a
improve the quality of multimedia communication.
DisengageRequest (DRQ) message to inform the
gatekeeper that a call is being terminated. The gatekeeper Multimedia call centers are another application for H.323.
replies with a confirm (DCF) or reject (DRJ) message. An H.323 call center provides a well-integrated
Alternatively, endpoints may unregister from the environment for Web access and other data/voice business
gatekeeper by sending an UnregisterRequest (URQ) situations. The call centers are used by banks for customer
message. The gatekeeper replies with an service, shops for extra retail outlets, etc. The call center
UnregisterConfirm (UCF) message or an can just be an H.323 terminal or an MCU, or it can be a
UnregisterReject (URJ) message. full featured endpoint with a gatekeeper, a gateway, and
MCU capability. The front end of the legacy call center
H.323 Deployment Obstacles may be a gateway, which allows installed systems to
operate with minimal disruption.
In order to achieve H.323 deployment in real networks,
limitations at both the network level and the client Another compelling application for H.323 is
platform level must be resolved at the H.323 client. The telecommuting. Telecommuters can attend meetings at
client should scale performance based on the available their companies, check their mail, or talk with someone at
bandwidth. Given the inconsistencies of networks with the company while on the road or at home.
best effort traffic, (i.e., no guaranteed Quality of Sevice
Finally, IP telephony is another area in which H.323 has
(QoS), the Internet), it is extremely important to provide
found a significant role; this will be covered in a later
mechanisms for fault tolerance and error resiliency at the
section.
client platform, if it is to be used on an unmanaged
network such as the Internet. At the network level, broad
connectivity, policy management, and security are H.323 Zone Management
considered the major issues in the deployment of a new Gatekeepers fulfill a required set of operational
technology such as H.323. Switched Circuit Network responsibilities and may offer a number of optional
connectivity is achieved in the H.323 context by using functions to entities within their zone. Before we describe
gateways for H.320, H.324, H.323, POTS, and other how zones are managed, we will review some of these
endpoints on other networks. Policy management is responsibilities and functions.
achieved using gatekeepers to provide call admission,
A gatekeeper acts as a monitor of all H.323 calls within its
authentication, and zone management. Deployment of
zone on the network. It has two main responsibilities: call
QoS protocols such as RSVP can also help with policy
approval and address resolution. An H.323 client that
management. The security of media streams is another
wants to place a call can do so with the assistance of the
important factor in the success of H.323 deployment,
gatekeeper. The gatekeeper provides the address
especially in unsecured environments such as the Internet.
resolution to the destination client. (This division of work
The platform consists of two main components: the is due to alias name registration procedures.) During this
operating system and the CPU. Many media compression address resolution phase, the gatekeeper may also make
algorithms are currently limited by the machine speed. permissioning decisions based upon available bandwidth.
(We expect this issue to improve as more powerful The gatekeeper can act as an administrative point on the
processors hit the market.) Moreover, popular desktop network for IT/IS managers to control H.323 traffic on
operating systems do not supply consistent real-time and off the network.
services. Multitasking can adversely affect the quality of
Strictly speaking, a gatekeeper zone is defined by what it
audio and video in an H.323 conference.
contains: it is defined by all of the endpoints, gateways,
and MC(U)s that are or will be registered with a
gatekeeper. Another way to describe a gatekeeper zone is
H.323 Applications to call it an “administrative domain,” although the formal
ITU-T recommendation text uses zone. An example of
A number of applications can take advantage of H.323
gatekeeper zones is given in Figure 5.
technology both in the corporate environment and in the
home-user environment. One obvious application is video



Some important aspects of zone coverage are summarized Control (MC). In other situations, an MCU is required; we
as follows: will explain why in the next section.
• Zones are defined by all H.323 devices registered to a The two broad models of multipoint, centralized and
single gatekeeper. decentralized, differ in their handling of real-time media
streams (audio and video). The centralized model operates
• Zone design may be independent of physical in the same fashion as other circuit-based conferencing
topology. (e.g., H.320 or telephony “bridges”). In this model, all of
• Each zone has only one gatekeeper. the audio and video is transmitted to a central MCU that
mixes the multiple audio streams, selects the
• Zone definition is implementation-specific. corresponding video stream, and re-transmits the result to
• Gatekeeper zones are logical in nature. all of the participants. Figure 6 illustrates this procedure.

• Network topology and administrative scope are both
factors in zone design.
• Resources such as gateways and proxies may affect A F
the partitioning of zones.

MCU
B E

Zone A Video
C
Audio
D

H.323 H.323
Gatekeeper
Centralized Multipoint

H.323
H.323 Figure 6: Centralized model

Note that the MCU is only ‘logically’ at the center of the
H.323
conference configuration; the physical topology may be
Gatekeeper very different. The operational model of the conference is
such that each endpoint is exchanging media control
H.323 signaling (H.245), audio, and video directly with the
Zone B MCU. In order to prevent echo, the MCU must provide
H.323
the current speaker with a custom audio mix that does not
contain the speaker’s own audio. The remaining
participants may all receive the same audio and video
media. In some implementations, the MCU may actually
Figure 5: Gatekeeper zones decode the video streams and combine them in a mixing
operation to provide continuous presence of all
Multipoint Conferences: Centralized Versus participants on the endpoint displays. The amount of
Decentralized processor speed required for this operation increases
significantly.
Multipoint conferences are defined as calls between three
or more parties. During these conferences, call control and The decentralized model shares common control
media operations can become considerably more complex characteristics with the centralized model, but the media
than during a simple point-to-point conference. The streams are handled differently. One of the participating
coordination and notification of participants entering and entities must be an MC, but it will typically be co-located
leaving a conference along with the marshalling of the with one of the endpoints. All of the H.245 connections
media streams require the presence of at least a Multipoint will have one end terminating at the MC just as with the



MCU in the centralized model. Whereas the MCU does level. The decentralized multipoint model does require
the media processing in the centralized model, the media that one of the participating entities contain an MC. If the
streams are sent and received by all participating entities endpoint that has the MC leaves the conference, the MC
on a peer-to-peer basis in the decentralized model. As must stay active or the conference is terminated. In order
Figure 7 demonstrates, the logical configuration for this is for this model to be bandwidth efficient, the underlying
a bus. There is no MCU to process the multiple streams; network should support multicast. If it doesn’t, each
each entity is responsible for its own audio mixing and endpoint can send multiple uni-cast streams to all others,
video selection. The media may be sent between all but this becomes increasingly inefficient with more than
entities utilizing either multicast, or a multiple uni-cast if four entities participating in a conference.
the underlying network does not support multicast.
H.323 Features for IP Telephony
The initial environment envisioned for H.323 was the
corporate network environment, primarily local area
networks. Wide Area Network (WAN) access was to be
gained by using gateways to H.320/ISDN. During the
implementation of Revision 1 of H.323, it became clear
that IP telephony was gaining popularity and relevance as
Video
various infrastructure elements were improved upon. A
Audio number of proprietary IP-based telephones were creating
A B C D E F many small islands that could not communicate with one
MC another. Recommendation H.323 provided a good basis
for establishing a universal IP voice and multimedia
Decentralized Multipoint
communication in larger, connected networks. With
Revision 2 of the Recommendation, new additions and
further extensions were added specifically to make it more
suitable for IP telephony. These changes will be described
Figure 7: Decentralized model in a later section.
By using Q.931 as its basis for establishing a connection,
There are a number of advantages and disadvantages to H.323 allows for relatively easy bridging to the public
both the centralized and the decentralized models. In the switched telephone networks (PSTN) and circuit-based
centralized model, the endpoints that participate in this phones. The required voice codec of G.711 also allows for
type of conference are not required to be as powerful as in easy connections to the legacy networks of telephones.
the decentralized model. Each endpoint has only to The uncompressed 64kb/sec stream can easily be
encode its locally produced media streams and decode the translated between digital and analog media. One of the
set sent by the MCU. The MCU may provide specialized, addressing formats provided in Recommendation H.323 is
or higher performance conferencing. The MCU’s ability the E.164 address. This is another ITU Recommendation
to provide custom-mixed media streams can allow that specifies standard telephone numbers (e.g., the digits
otherwise constrained endpoints to participate in 0-9, * and #). These addresses, which ultimately map onto
conferences. For example, if an endpoint is connected to the IP addresses for the H.323 endpoints, allow regular
the conference by a low-bitrate connection, it may only telephones to ‘dial’ them. Gatekeepers provide the final
have the capacity to transmit and receive one media important element for IP telephony. Gatekeepers supply
stream. If this same endpoint were connected to a the ability to have integrated directory and routing
decentralized conference, it would have no ability to functions within the course of the call setup. These
detect which speaker was the focus when the focus operations are important for real-time voice or video when
changed. In most situations, the MCU will choose the resources must be balanced, and points of connectivity are
largest common set of attributes within which to operate highly dynamic. The gatekeeper functions, which provide
the conference. This may lead to a conference that is call permissioning and bandwidth control, enable load
operated in a least common denominator mode, such as monitoring, provisioning, and ultimately, commercial-
QCIF, rather than CIF or lower video resolution. grade IP telephony service.
By definition, the decentralized multipoint model does not
require the presence of an MCU, a potentially expensive
Interoperability
and limited resource. It allows for disproportionate Recommendation H.323 comprises a number of
processing at the endpoints with each running at its own interrelated documents and sub-recommendations.



Therefore, customers are faced with a number of options, Porting of the existing Q.931 code to the packet
which they may or may not implement. These design environment provided mis-matches. During the initial
choices, in conjunction with differing interpretations of course of development, there were a small number of
the required elements, can lead to implementations that defects in the Recommendation that were discovered by
cannot interoperate. Stated more simply, compliance with the implementers. These defects were recorded in a
the recommendations does not always imply document called the Implementer’s Guide and eventually
interoperability. The complexity and flexibility of H.323 corrected in the revision of the base Recommendation. In
essentially requires that implementations be tested addition to the low-level control protocol interoperability,
together to ensure interoperability. the media stream is another potentially problematic area.
Recommendation H.245 allows the receiver to specify the
To this end, the International Multimedia Technical
maximum bitrate that may be sent, but currently there is
Consortium (IMTC) has focussed largely on this area of
no way in which to specify the maximum Real Time
interoperability. The testing events sponsored by this
Protocol (RTP) packet size. This can lead to
consortium have provided a venue for more than thirty
interoperability problems if a receiver implementation
companies to participate in “bake-offs” to test the
makes assumptions about the buffering required and then
interoperability of their product. The methodology that is
cannot decode the packet. Although H.323 specifies
followed provides for a gradual testing of component
G.711 and H.261 as its baseline codecs (audio and video
protocol stacks that then leads to higher level end-end
respectively), a large number of initial H.323
scenario testing. Figure 8 illustrates the grouping of the
implementations were targeted for low-bitrate connectivity
components. Each horizontal level provides increasing
where these codecs could not operate. The result of this
H.323 functionality, and each vertical group indicates a
was that the audio codec, G.723.1, was chosen due to its
specific stack function that is required.
low (5.3-6.4kb/s) data rate. This low bitrate allowed
Much of the mass interoperability testing has occurred in H.263 video to operate in an acceptable fashion across a
face-to-face events. In theory, the H.323 protocol should 33.6kb connection.
allow for testing from remote sites across the Internet; in
Implementation choices that are made as a result of
practice, however, remote testing has not turned out to be
operating environments elevate the interoperability issues
useful. The lack of a controlled environment makes
to the next level. At an operational level, endpoint
distinguishing interoperability issues from simple network
implementations may select to support an asymmetrical
problems extremely difficult.
media model for optimal performance; other
Examples for initial interoperability problems encountered implementations may not. The ability to enter extended
during some of the testing events include misaligned bit information at the user interface may determine whether
field and byte-ordering issues. These problems are an endpoint can operate with certain gateways or H.323
common in the early development stage of protocols. For proxies. Gatekeepers may or may not provide intelligent
example, there are a number of adaptations that are zone control, which allows them to operate in a network
specified in the Q.931 protocol used by H.323, which connected to other activated gatekeepers. With the
make the Protocol Data Units (PDUs) slightly different addition of a range of security options to H.323, strict
from the protocol messages used in the circuit world (such policy constraints may prevent H.323 entities from
as ISDN). interoperating while still complying with all the
recommendations.



Group Group Group
Group Group Group Group Group Group
A B D
C E F G H I
Audio Video RTP/
Layer 0 TPKT RAS MC T120
Codec Codec RTCP

Audio Video H.245
Q.931
Codec Codec Control
Layer 1 RTP/ RTP/
TPKT TPKT
RTCP RTCP

Group J Group K

H.245 H.245
Q.931 Q.931 RAS
Layer 2 Control Control
TPKT TPKT

Group L

Audio Video H.245
Layer 3 Q.931
Codec Codec Control
RTP/RTCP TPKT

Group M Group N

Audio Video H.245 Audio Video H.245
Q.931 RAS Q.931 T120
Layer 4 Codec Codec Control Codec Codec Control
RTP/RTCP TPKT RTP/RTCP TPKT

Figure 8: Testing matrix1
this allows a ‘line hunting’ mode of operation. Keep alive
New Features types of messages called Request In Progress (RIP) can
Revision 2 (1998) of Recommendation H.323 contains a stop premature retries on lengthy operations.
number of improvements for IP Telephony, among other
The call setup has a new method called FastStart that
areas. A new Recommendation (H.235) was developed to
establishes bi-directional media in one round trip time of
provide a full security framework for H.323 and other
messages (discounting the establishment of the actual TCP
multimedia systems. It may provide the services of
connection). Figure 9 shows FastStart messages
Authentication (which can be used for authorization),
exchanged. The OpenLogicalChannels messages that
Privacy, and Integrity. The system can utilize underlying
occur after H.245 is established in the regular case are
security protocols such as Internet Protocol Security
piggy-backed on the Setup-Connect exchange. This
(IPSEC) or Transport Layer Security (TLS) as established
facility allows instant audio connection that resembles the
in the IETF. A number of new features and options were
regular phone call model as opposed to the standard
introduced regarding interactions with gatekeepers.
lengthy H.323 startup procedures.
During the process of discovering gatekeepers, endpoints
may actually receive a number of gatekeeper addresses to
utilize. This redundancy in the protocol will eliminate the
single point of failure if a gatekeeper becomes
inoperative. When endpoints register their current address
with a gatekeeper, they may specify multiple addresses;

1
In this figure TPKT refers to the layer that segments a TCP stream into separate message ‘packets’ to be delivered to the
H.323 application.



Members may participate in the conference by joining the
Caller Callee panel. They may also leave as they wish. This H.323 panel
“A” “B” is analogous to a panel on a stage in a large auditorium.
TCP connection
Occasionally participants might get up from the audience
(H245 Address)
SETUP (RTCP & RTP addresses) and join the panel and others might leave the panel and
join the audience. This movement of participants occurs
using the standard Q.931 Invite and Join signaling as
CONNECT (H245 Address)
(RTCP & RTP addresses)
described in the H.225.0 document.

RTP stream
RTP stream Conclusion
RTCP stream The H.323 system and its associated recommendations
provide a useful and flexible system for multimedia
TCP communication. The factors that allow the protocols to
connection
easily bridge data and voice networks also make H.323
H.245
Messages
scalable. The ability of most, if not all of the system, to
operate on a general-use platform such as a personal
ReleaseComplete
computer allows the H.323 system to scale with
underlying processing power. As the processor’s speed
budget increases, the H.323 system can provide a better
end-user experience. The dynamic exchange of
Figure 9: FastStart capabilities allows the communications to change if
Many new explicit address types have been added. needed during a call and adapt to any environmental or
including RFC822 (e-mail) formatted names and URLs, endpoint constraints.
which network-based users find more familiar and easier Recommendation H.323 has become the single standards-
to remember. Each endpoint may register under a number based solution for a complete array of communication
of these aliases. New unique call identifiers were added to systems from simple point-to-point telephony to a rich
better track calls as they traverse multiple network nodes multimedia conference with data sharing. Through
and topologies. A number of new features were added to continued effort by the ITU-T study group,
H.245 also, which allow for a much richer set of media Recommendation H.323 continues to evolve and adapt to
possibilities. In addition to layered codecs, which divide new situations. Many of the real world difficulties in
up the video stream into additive layers of detail, Quality utilizing H.323 come about from infrastructure issues or
of Service parameters (e.g., RSVP) may now be signaled other problems that are being resolved. Globally
when opening up media streams. Finally, the ability to coordinated addressing and consistent QoS are two areas
pass the media on other transports such as ATM (while where we expect to see great improvements in the future.
keeping the rest of call and media control on IP) have Some of these improvements will be facilitated by
been added. expected higher level communications between
In addition, a new family of recommendations has been gatekeepers and gateways as their interaction is
developed to set up a framework for supplementary standardized.
services. The H.450.1, H.450.2, and H.450.3 are loosely
based on the Q-signals (QSIG) protocol, which is used by References
PBX and switch vendors. The initial services that have
been defined are call transfer and call forward. Both of
these features are a peer-peer option requiring no specific
help from a centralized network entity (such as the PBX in [1] ITU-T Recommendation G.711 (1988)“Pulse Code
Modulation (PCM) of Voice Frequencies.”
the circuit world). A number of standardized
supplementary services are expected to be developed in
the future. [2] ITU-T Recommendation G.723.1 (1996) “Dual Rate
Speech Coders for Multimedia Communication
Along with the new features, the latest development of Transmitting at 5.3 & 6.3 kb/s.”
H.323 also includes an expanded topology model in the
H.332 document. Recommendation H.332 allows an
H.323 conference model to expand to literally thousands [3] ITU-T Recommendation H.261 (1993) “Video Codec
of participants. The tightly controlled H.323 ‘panel’ is for Audiovisual Services at p X 64 kb/s.”
surrounded by a very large number of ‘listeners.’



[4] ITU-T Recommendation H.263 (1996) “Video Outside of Intel, Mr. Toga develops standards and
Coding for Low Bit-rate Communication.” standards-based products within ITU-T, IETF, and IMTC.
He is also involved in the following related activities:
[5] ITU-T Recommendation H.323 (1998) “Packet Based Editor of ITU-T H.323 Implementors Guide
Multimedia Communications Systems.” H.235 Security Standard
Chair of IMTC “Packet Networking Activity Group”
[6] ITU-T Recommendation H.225.0 (1998) "Call Chair of H.323 Interoperability Group
Signaling Protocols and Media Stream Packetization He has also written numerous articles for trade magazines.
for Packet Based Multimedia Communications His e-mail is jim.toga@intel.com.
Systems."

[7] ITU-T Recommendation H.245 (1998) "Control Hani ElGebaly is the project technical lead for the
Protocol for Multimedia Communication." H.323/H.324 protocols development team within the
Conferencing Products Division of Intel Architecture
Labs. He received an M.Sc. in Computer Science from the
[8] ITU-T Recommendation H.246 (1998) “Interworking
of H-Series Multimedia Terminals.” University of Saskatchewan, Canada, and a B.Sc. in
Electrical Engineering from Cairo University, Egypt. Mr.
ElGebaly is currently pursuing a Ph.D. with the University
[9] ITU-T Recommendation H.235 (1998) “Security and of Victoria, Canada. His primary areas of interest include
Encryption of H series (H.323 and other H.245 based) multimedia conferencing protocols, embedded
Multimedia Terminals.”
programming, fault tolerant systems, and computer
architecture. His e-mail is hani.el-gebaly@intel.com.
[10] ITU-T Recommendation H.332 (1998) “Loosely
Coupled H.323 Conferencing.”

[11] ITU-T Recommendation H.450.1 (1998) “Generic
Functional Protocol.”

[12] ITU-T Recommendation Q.931 (1993) “Digital
Subscriber Signaling System No. 1 (DSS 1)-ISDN
User-Network Interface Layer 3 Specification for
Basic Call Control.”

Authors’ Biographies
Jim Toga holds a B.Sc in Chemistry from Tufts University
and a M.Sc in Computer Science from Northeastern
University. Before joining Intel, he was the principal
engineer on StreetTalk∗ Directory with Banyan Systems
where he designed and developed the Yellow Pages
service. Presently, he is a senior staff software architect
for the Standards and Architecture Group in the Intel
Architecture Labs. He coordinates product groups giving
guidance on architecture and standards. His primary tasks
are H.323/Internet Telephony, Directory, and real-time
security issues.

∗
All other trademarks are the property of their respective
owners.


Demystifying Multimedia Conferencing Over the Internet Using ...

More Related Content

What's hot (19)

Similar to Demystifying Multimedia Conferencing Over the Internet Using ... (20)

More from Videoguy (20)

Demystifying Multimedia Conferencing Over the Internet Using ...