To many people, the very term H.323 has become a synonym for those of IP telephony and multimedia. Most products that implement the multimedia technologies—starting with PCs and extending to gatekeepers and media gateways—have implemented the set of standards commonly referred to as H.323. For this reason, the convergence products refer to H.323 systematically.
ITU-T Recommendation H.323 specifies an umbrella standard for real-time multimedia communications over packet networks. The standard refers extensively to other ITU-T Recommendations for certain specifics while defining the relevant system components and how these components interact with each other. Most important among the referred ITU-T Recommendations are H.225.0 and H.245, which specify the protocols used between the H.323 components for call and media control.
H.323 was first developed by the ITU-T Study Group 15. As a result of the ITU-T reorganization in 1997 to consolidate multimedia-related work spread across multiple study groups, the work has been moved to newly formed Study Group 16 since then.
The initial H.323 Recommendation, entitled “Visual Telephone Systems and Equipment for Local Area Networks (LANs) that Do Not Provide Guaranteed Quality of Service (QoS),” was approved in 1996. As the title suggests, the Recommendation had a specific scope concerning a particular type of packet networks. Over the next two years, in response to the growing public interest in the Internet, the scope of H.323 was considerably extended to cover metropolitan area networks, intranetworks, and internetworks. Consequently, a new version (publicly know as version 2) of ITU-T Recommendation H.323 was approved in 1998. The Recommendation also received a new title—“Packet-Based Multimedia Communications Systems.” The rest of the section describes the major aspects of this Recommendation.
H.323 draws on the constructs of calls and channels for modeling a communication session. A call is defined as point-to-point communications between two H.323 endpoints. (An endpoint can both call and be called. It can also send or receive—or both send and receive—media streams.) A typical call begins with a call initiation attempt issued by the calling endpoint to the called endpoint and ends with a termination request from either endpoint. Specifically, H.323 divides the life cycle of a call into the following phases:
§ Phase A. Call setup (that is, initiation of a call).
§ Phase B. Initial communication and capability exchange (that is, determination of capabilities supported by endpoints and the endpoint to serve as the master at times of conflicts).
§ Phase C. Establishment of audiovisual communication.
§ Phase D. Call services (for example, increase of bandwidth and addition of conferencing parties during the call).
§ Phase E. Call termination.
Channels are used as building blocks for calls: A collection of channels between two endpoints makes up a call. Conceptually, channels (normally unidirectional) are used to transmit signaling messages and media streams. Channels can be either reliable or unreliable. Reliable channels (for example, TCP connections) are used to carry signaling messages (for example, H.225.0 and H.245 messages), and unreliable channels (for example, RTP sessions) are used to carry media streams.
Figure 1depicts an H.323 system consisting of typical H.323 components, including the terminal, multipoint control unit (MCU),gateway,and gatekeeper. With the exception of the gatekeeper, these components can serve as endpoints. Before getting into the detail of the components, we should point out an important H.323 requirement: Its components must interwork with multimedia communications terminals for other types of network connections.
Figure 1: H.323 system.
Source: ITU-T Recommendation H.323: Figure 1.
This interworking consideration resulted in adoption of the Q.931 access protocol as the basis for the protocols of Recommendations H.225.0 and H.245.
A terminal provides the real-time audio and, optionally, video and data communications capabilities for the direct use of users in a point-to-point call or multipoint conference. Figure 2 depicts the internals of a terminal. The H.323 Recommendation prescribes that a terminal include an audio codec unit, system control unit, H.225 layer, and network interface. It also defines each of those elements, with the exception of the network interface, which is noted as being out of the Recommendation’s scope. We briefly summarize the defined elements here.
Figure 2: H.323 terminal.
Source: ITU-T Recommendation H.323: Figure 4.
As shown in Figure 2, several audio codec standards are specified in H.323. In particular, the G.711 codec must be supported by any H.323 audio device. This support is prescribed for a good reason: G.711 is the coding scheme used in the PSTN for digitizing voice; it is considered the benchmark voice codec. Other coding schemes (such as G.723.1 and G.729) are optional, however.
The system control unit performs signaling operations that concern the communications between the terminal and other endpoints or gatekeepers. It provides for call control, capability exchange, logical channel control, and signaling of command and indications. As shown in Figure 2, these functions must use the mechanisms defined in Recommendations H.225.0 and H.245, which we discuss later in this section.
Finally comes the H.225.0 layer, which packs the media (a combination of audio, data, and video) streams and control messages for transmission and unpacks the received packets through the network interface. As the name of the layer implies, the specifics of packaging are defined in H.225.0. An important prescription is that RTP must be used to carry audio and video streams over unreliable logical channels. Another is that there must be no fragmentation of H.225.0 and H.245 signaling or control messages across transport packet data units.
The gateway provides the functions that are required for interworking between an H.323 terminal on a packet network and an ISDN terminal or POTS telephone of the PSTN (as depicted in Figure 1). Through gateways, IP-based telephones are connected with a billion or so PSTN telephones. Specifically, the gateway provides a conversion function for the transmission format, as well as call control and logical channel control messages. It may have the characteristics of a terminal or an MCU, but the choice (terminal or MCU) is left to the manufacturer. The gateway communicates with other H.323 endpoints or gatekeepers using the same means as H.323 terminals. Its communications with PSTN terminals are beyond the scope of the H.323 Recommendation.
Because it has an important role in IP telephony, the gateway issue is tackled by various standards bodies. Several new interfaces and protocols are being developed to support large-scale deployment of IP telephony services.
The gatekeeper is optional in an H.323 system. Yet, when present, it plays an important role—as its name implies. The gatekeeper provides the following services:
§ Address translation. Translates an alias address (for example, a telephone number or an e-mail address) to a network address (for example, an IP address). This function is particularly important in scenarios where a telephone on the PSTN is attempting to call a PC on an IP network. It is also important when determining the location of a gateway to the PSTN.
§ Admission control. Authorizes network access using the registration, admission, and status (RAS) messages defined in H.225.0. The rules or policies for authorization are unspecified, however, so that they can be customized for a given situation. Examples: (1) admit all requests or (2) admit only requests from subscribers.
§ Bandwidth control. Handles bandwidth change requests from an endpoint. The requests and responses must use the corresponding RAS messages. The H.323 Recommendation leaves open how bandwidth requests should be handled.
§ Zone management. Manages a zone, the collection of all endpoints registered with and managed by one and only one gatekeeper. The gatekeeper, in turn, must support the registration and management of all endpoints registered in the zone. The H.323 Recommendation does not specify the rules governing the geographic or logical composition of a zone, nor does it specify how two gatekeepers communicate with each other across zones. Related methods for communications between administrative domains can be found in a separate Recommendation—H.225.0 Annex G.
In addition, the gatekeeper may optionally provide the following services:
§ Call control signaling. Call signaling messages between endpoints may be passed with or without the involvement of the gatekeeper, as illustrated in Figures 3 and 4. If the gatekeeper is involved (this case is called gatekeeper-routed call signaling), the call signaling messages are routed through the gatekeeper; otherwise, they are passed directly between the endpoints (this case is known as direct endpoint call signaling). The choice of the call signaling method belongs to the gatekeeper. In the response to an endpoint’s admission request, the gatekeeper must indicate the call signaling method. Note that gatekeeper-routed call signaling is of particular relevance to service providers and enterprise network managers. The routing method supports collection of calling activities at a central place and provides an entry point for additional policy enforcement and service processing during a call, thereby enabling advanced resource management and services.
Figure 3: Gatekeeper-routed call signaling.
Source: ITU-T Recommendation H.323: Figure 9.
Figure 4: Direct endpoint call signaling.
Source: ITU-T Recommendation H.323: Figure 10.
§ Call authorization. The authorization policy (unspecified by the Recommendation) can be based on, for example, restricted access to a gateway, time of day, type of service subscription, and bandwidth availability.
§ Bandwidth management. A call request or bandwidth request may be rejected because of bandwidth limitations. Theoretically, the gatekeeper could also instruct an ongoing call to reduce its bandwidth usage. The criteria for determining whether there is bandwidth available are outside the scope of H.323.
§ Call management. The gatekeeper may maintain a list of ongoing H.323 calls and provide call management based on the information. For example, for a call to a busy terminal, the gatekeeper can redirect the call or simply not make an attempt to establish the call.
The MCU provides control functions to support conferences of three or more endpoints. It manages conference resources and negotiates with the endpoints the modes of communications. Optionally, the MCU may also provide functions, such as mixing and switching, for processing the media streams it receives before distributing them to endpoints. Mixing is the operation for combining multiple-input media streams into one single output stream; switching is the operation for selecting an input media stream for output to an endpoint. The MCU uses H.245 messages and procedures for establishing multipoint communications between the endpoints.
H.323 also stipulates the procedures the communicating endpoints should follow when exchanging signaling messages throughout a call. It specifies the sequencing of relevant H.225.0 (including RAS) and H.245 messages for every phase of a call in various scenarios. In particular, the call setup procedures cover the following scenarios concerning the role of gatekeepers:
§ Neither of the endpoints is registered with a gatekeeper
§ Each endpoint is registered with a different gatekeeper
§ Only the calling endpoint is registered with a gatekeeper
§ Only the called endpoint is registered with a gatekeeper
Note that the call setup procedures support a signaling method known as fast connect, which allows the establishment of media channels without the use of H.245 signaling. As a result, media channels can be opened for sending and receiving media information with as few as one round-trip signaling message exchange.
No comments:
Post a Comment