Tuesday, August 10, 2010

Multipoint Conferencing Units (MCUs)

MCUs provide a means of implementing the multimedia conferencing services. The products come in several grades, depending on the number of ports and features they support. On the switching side, MCUs can be connected to PBXs, central offices, or toll switches via trunks; the rest of the connections are to the endpoints, a gatekeeper, or other MCUs for cascading. MCUs may also provide a connection to an external computer (server) that actually runs the conferencing application software. In the latter case, more than one MCU is typically connected to the server.

MCU hardware arrives in cabinets hosting multiple circuit packs with video and audio codecs that are connected internally via a fast switch. In essence, MCUs are specialized multimedia gateways augmented by conferencing capabilities. In high-end products, the packs are equipped with clearly visible green, yellow, and red light-emitting diodes (LEDs) that indicate real-time status. The LEDs can be easily removed (via the turn of a single screw) and replaced without powering down the whole system.

At the time of this writing, MCUs support from 4 to 20 video ports in low-end products, up to 42 in medium-end products, and up to 64 in high-end products. Other parameters include the number and supported types of audio ports, the total number of conference participants, the total number of conferences per MCU, and the number of MCUs that can be connected for cascading—an important indicator of scalability. Overall, one large MCU is typically less expensive than several smaller MCUs with the same total capacity; the maintenance and operations costs are also lower if only one MCU is employed.

High-end MCUs also support bonding (that is, the capability of aggregating independent 56/64-kbps calls) to achieve high-bandwidth conferences. For example, ten 64-kbps calls from one endpoint can be bonded for a 640-kbps conference. Currently, high-end products support up to 24 T.120 multipoint data conferencing ports, H.320 ports, and active conferences per MCU, and can interconnect with up to 50 MCUs for cascading.

Support of video, audio, and data conferencing standards is essential, although high-end products invariably support proprietary codecs (and provide transcoding when necessary).

Some existing high-end products support all of the following features:

  • Chair control. The conference chairperson can request or relinquish control of the conference, choose the broadcaster, and drop a site or a conference.
  • Continuous presence. Several images (up to 25 in high-end products) can be maintained on a participant’s screen simultaneously. For example, a picture composed of images taken from several participating sites is composed and maintained at video rates from 56 to 768 kbps and 30 frames per second. Another example is display of panoramic view (derived by maintaining several simultaneous images of a conference site).
  • Broadcast with autoscan. Participants can see the speaker all the time, but the speaker sees the participants on a rotating basis.
  • Voice activation. The MCU automatically switches the video display when the speaker changes. It is important to test voice activation, because unexpected switching of the display can occur due to background or other accidental noise. High-quality products switch the display only in reaction to the human voice.
  • Full-duplex audio. This is present only in high-end products because it requires rather sophisticated software control of the DSPs. Every conference participant can talk and hear what others are saying at the same time. Full-duplex audio is important for highly interactive conferences, especially to those who are participating as audio only.
  • Presentation mode. Participants can see the presenter at all times, but the presenter can see only the person who is asking questions.
  • Terminal naming. Conference participants are identified by strings of characters associated with their terminals (typically the names of participants, but sometimes titles or geographic locations). The name is displayed when a person associated with the terminal speaks. For terminal naming to work, the terminals must support the H.243 standard.
  • Speed match. Participants can join the conference at different transfer rates (varying from 56 to 768 kbps). In particular, low-speed endpoints can still be admitted to higher-speed conferences for audio only.
  • Still image transfer. Still images can be transmitted at 4 times the quality of moving images. The images are transmitted using the H.320 video channel (rather than data channels, which are endpoint dependent); the endpoints must support Annex D of ITU-T Recommendation H.261 for still image transfer to work.
  • Data conferencing. Participants can share data in real time. Spreadsheet applications, word processors, and file transfer applications are typically supported.
  • Universal control. The conference parameters (for example, composition of the screen) can be controlled by a device as primitive as a touch-tone telephone or as sophisticated as a highly specialized real-time graphical user interface application (often part of a product offering) running on a PC.
  • Conference mobility. Useful for trouble isolation as well as some advanced services, such as meet and greet. Conference participants can be temporarily moved from one conference to another.
  • Audio mode selection. The codec to be used for the conference can be selected at the time of reservation.
  • Dial-in (meet me) conferencing. Conference participants dial into the conference. High-end products support a variation called one-number access, which allows participants to dial the same number (regardless of the number of cascaded MCUs) to enter the conference.


  • Dial-out conferencing. The MCU itself dials the participants of a conference. Dial-out conferencing is essential when the conference participants are spread around the world and the MCU is located in the United States, since international calls tend to be cheaper in the United States than in rest of the world. Another important benefit of dial-out conferencing is that it automatically results in one bill. The same billing can be achieved with dial-in conferences.


  • Web-based reservations. A conference host can schedule a conference by accessing a Web page on the enterprise intranet. Once the conference is scheduled, records can be updated by the person who has scheduled it. In some implementations, a password is maintained for each conference schedule. This password is made known only to conference participants, who are prompted for it when they join the conference.


The preceding features are used by conference participants. There are features pertinent to system management that do not affect conference participants directly. Faultless system management and fast troubleshooting are essential to the perception of high-quality service, and so are its enabling features, as follows:

  • Active serviceability. Available only in high-end products, this feature allows the hot-swapping of port circuit packs without any disruption to active conferences. Using hot-swapping, you can change a circuit packet without powering off and on, so it is not necessary to reboot the system. Furthermore, advanced products support automatic self-diagnostics and self-monitoring as well as customer-programmable fault isolation procedures. For example, ISDN interfaces can be programmed to filter and display specific ISDN messages based on trigger conditions (such as channel number or calling or called party numbers) and keep a trace of them as well as H.320 signaling exchanges. Error and event logs are time-stamped and kept for real-time and postconference analysis. Service administrators (who should be able to log in remotely) are provided with a large number of commands for testing the resources and modifying the runtime parameters if necessary. The administrators can also decide which conditions should trigger problem alerts as well as choose a particular form of alert. Typical available forms of alerts include dialing out the alarm collection center, lighting an LED on the maintenance telephone, and a range of other audible and visible alerts communicated to specialized network management centers.


  • Conference monitoring. Systems administrators can request that a snapshot of any conference in progress be provided on the MCU console screen. With some products, the administrator can further request that this snapshot be updated at a given frequency (for example, every 30 seconds). Another feature is roll call, which allows the conference administrator to view and hear every endpoint participating in the conference.


  • Service flag. Reservation agents can enhance endpoint compliance. Based on the historical information of an endpoint, the agents can manually activate a flag that would allow that endpoint to enter a conference with only a subset of supported services. The subset is determined on the basis of historical compliance information; access to services that require stronger compliance than the endpoint is capable of would be prevented.


  • Software control of dial-in numbers. A single number can be used at different times of day for different dial-in capacities (a bonded call, an H.320 call, and so on), as requested by conference reservations. (The alternative, still employed in some products that maintain a static association of hardware to the dialed number, is manual reconfiguration of boards in the cabinet.) The dynamic association significantly simplifies maintenance of MCUs and at the same time reduces the total number of network addresses required and improves the reliability of the system. To appreciate the network address reduction point, note that the dynamic association of a number with a particular hardware resource actually takes place only at the start of a call. Thus, the resources are pooled and acquired by hunting. Consequently, a malfunctioning resource has no effect on service until all good resources have been used. (With static association, however, a malfunctioning board simply prevents calls to the number associated with it, so the number may not be used.)


  • Cascading. Multiple MCUs can be linked to create a single conference. This allows a significant increase in the number of endpoints on one conference when necessary. Naturally, the maximum number of MCUs that can be cascaded is an indication of the scalability of a particular make of MCU. High-end products also offer many features that contribute to a secure conference. H.320 password protection is almost invariably supported; audio conferencing passwords can be entered by using a telephone’s dial buttons.

No comments:

Post a Comment