INTRODUCTION TO H.323 DIGITAL VIDEO STANDARD

When this standard is built into applications, it passes audio, still-pictures and video over and across:

• existing LANs - Ethernet
• existing WAN
• the Internet
• ISDN (technically although the H.320 is more adapt)
• and POTS plain old telephone systems)

H.323 can be deployed wherever Internet Protocol (IP) is supported, regardless of physical network topology or protocol which includes Ethernet, ATM, FDDI, ISDN, T1, etc. The H.323 standard is both hardware and operating system independent, which means products can be manufactured by many different vendors for use in a variety of environments.

H.323 defines four major components and how they interact with each other, as well as how they interact with a circuit-switched network (e.g. H.320 conferencing systems). These four components are:

• endpoints
• gatekeepers
• gateways, and
• multipoint control units (MCUs).

H.323 also defines two modes for conferencing:

• point-to-point and
• multipoint.

Endpoints are the clients in an H.323 network. They are typically video conferencing, audio conferencing, or other multimedia systems implemented by end users to communicate in real time. The H.323 standard requires that every endpoint support certain functions and codecs that have been previously defined by the ITU. Endpoints are required to support the following:

• G.711 - An ITU standardfor speech compression
• H.245 - A protocol for controlling media between H.323 endpoints
• Q.931 - A signaling protocol for establishing and terminating call
• Registration/Admissions/Status (RAS) Channel - A data stream used to communicate with a gatekeeper (not used if a gatekeeper is not present)
• Real-Time Protocol/Real-Time Control Protocol (RTP/RTCP) - IETF application-level protocol for carrying packetized real-time media on IP networks

Optionally, H.323 endpoints may also support video, additional audio codecs, T.120 data conferencing, multicast, and Quality of Service (QoS). If an endpoint supports video, the standard requires a minimum of H.261 QCIF (Quarter Common Interchange Format - 176 x 144 pixels) support. Support for other video formats is optional.

The fact that there are many optional features for an endpoint creates significant challenges for H.323 interoperability. For example, an H.323 endpoint that supports only the basic requirements is "standards compliant;" however, when connecting with another compliant endpoint that supports optional video and data sharing, the conference will be limited to basic H.323 functionality. To the end user, the conference will be limited to G.711 audio and no video or data. In other words, they are limited to the same functionality as a standard telephone call.

A gatekeeper provides call authorization for both accepting and placing calls in its "zone." A zone consists of H.323 endpoints, H.323- to-H.320 gateways, H.323 proxies and H.323 multipoint units that are configured to be under the gatekeeper's control. A gatekeeper also allocates bandwidth and provides address translation between aliases and IP addresses. Each zone can have only one active gatekeeper, but there can be many zones, and therefore many gatekeepers within an organization. Gatekeeper support on an H.323 network is optional; however, if a gatekeeper is used, all endpoints on the network are required to make use of it. Gatekeepers are not required but may support bandwidth allocation during a call or other supplemental services like call accounting, routing, and network management.

An H.323 gateway provides connectivity between H.323 endpoints and other types of endpoints. A gateway allows an ISDN (H.320), POTS (H.324), or analog telephone system to call into an H.323 conference, and vice versa. The H.323 standard requires that gateways provide call signaling support, control channel messages, multiplexing, audio compression, and audio transcoding. The H.323 standard does not define the number of ports a gateway must support or the type of physical interface a gateway should have.

The H.323 standard defines two conferencing modes: point-to-point and multipoint. Point-to-point conferencing is simply two endpoints conferencing together. When more than two endpoints want to hold a conference, a Multipoint Control Unit, or MCU, is required. The MCU acts like another H.323 endpoint on the network and controls all aspects of the conference.

A Multipoint Control Unit (MCU), referred to as a Multipoint Control Server (MCS) by VideoServer, Inc., enables three or more H.323 endpoints to conference together. The MCU mixes all the audio coming from the endpoints and switches video based on some criteria, such as the loudest person speaking. H.323 MCUs are available in both hardware and software-only versions. A software-only implementation runs as an application on network servers, and the number of conference participants is a function of the server's MIPs.