Focus: IP Multicast
Multicast methods

A look at how IP multicasting can be deployed for e-learning and other media-based applications. by Dr Seamus Phan

For most time-strapped enterprises, it is no longer practical to try to gather hundreds of employees within a single classroom to train them.

Enter e-learning, which has become an integral part of any corporate or industrial training program. Besides allowing distance learning, e-learning also has the benefit of allowing employees to learn at their own pace, without conforming to a fixed schedule designed by traditional classroom training.

Casting
One of the most common forms of media broadcasting is unicasting, where multimedia content is streamed from one node to another. The most common application of unicasting is audio and videoconferencing. In the case of audio and videoconferencing, it is bidirectional unicasting, since traffic travels from one node to another in both directions. Unicasting is less bandwidth hungry, depending on the level of compression for the multimedia content. The more compression there is in the audio and video content, the less bandwidth consumption there will be.

On the other hand, multicasting is where multimedia content is streamed from one node to many downstream nodes. In multicasting, it is often unidirectional, where traffic travels from the singular broadcasting node to the downstream nodes, and usually no traffic travels back.

Multicasting and unicasting can be combined in the same classroom. For example, a trainer can use bidirectional unicasting to communicate with individual participants to communicate private or sensitive ideas, while multicasting is used by the facilitator to communicate pervasive and generic program content to every participant.

Multicasting used to be exorbitantly expensive, but not anymore. These days, you can have a full-fledged broadcasting and production studio for under US$5000, which will allow you to deliver on-demand video content, as well as live QuickTime streams. For example, here is a short breakdown if you are using the Mac platform:

• iMac for authoring (US$1399)
• iMac for server use (US$1399)
• MiniDV camcorder (US$600)
• Mac OS X Server (US$999)
• Mac OS X Client (US$129)
• iMovie (US$49)
• Sorenson 3 Broadcaster (US$199)

Other incidental expenditure may be necessary if you want a more professional feel.

For example, you may require a sturdy tripod with lockable wheels, studio lighting (known as "red heads" with daylight filter), softbox lights (for diffused lighting), reflector screens (to soften harsh shadows cast by lights), bluescreen fabric for chromakey effects, unidirectional or omnidirectional microphones, lapel microphones, studio headphones, studio TV monitors, DVcam recording and playback decks, Steadicam (for stabilising handheld camera work), and so on.

These professional equipment will obviously push your initial investment sky high, and may not be necessary for internal training. But bear in mind though, if you intend to deliver multicasting to your shareholders and the media in corporate communication initiatives as well as internal training audiences, then these additional equipment may be necessary to deliver more credible and professional results.

Sound
In the world of broadcasting, even in the Web-based environment, the key to success is not so much the visual quality of video, but the audio quality. Humans pick up audio defects much more than visual defects, with a greater inherent tolerance for poorer video than audio.

This means that you have to compensate for bandwidth utilization and participant tolerance, with a bias towards audio rather than video. Some of the better audio compression such as QDesign (www.qdesign.com) works across both Mac and Windows platforms, and can be highly compressed while retaining audio quality. If you cannot or choose not to use QDesign, you may use more basic compression such as ALaw or µLaw compression, which are more compatible, but will generate larger file sizes.

You should invest in high-quality microphones, rather than use cheap US$10 unpowered microphones. Some of the best microphones are made by Shure (www.shure.com), and their powered microphones will be able to record a wider and deeper acoustic range. For discreet recording, choose the Subminiature Lavalier Microphone, which is an omnidirectional microphone. If you are narrating segments, choose the Headworn Microphone, which has noise canceling features.

Since audio is typically overlaid on video, you should record all narration as mono tracks. If you cannot record onto mono physically, you may merge the stereo tracks into mono via software, or simply delete one of the stereo tracks. This will then allow you to easily add audio dubs on your video tracks later.

QuickTime (www.apple.com/ quicktime) has wider audio response and sound level fidelity, which means that it more closely approximates what you hear. If you are using Microsoft Windows Media (www.microsoft.com/windowsmedia), it will be quite close to QuickTime quality as well, although the sound may become slightly higher-pitched, and more "tinny". If you are using RealAudio (www.real.com), then be aware that the sound level will be reduced (i.e. the volume becomes lower) and you may need to compensate via software.

Network good enough?
Of course, any video or audio broadcast can look great if run across an Intranet on a 100Base-T infrastructure.

However, most learning or corporate broadcast initiatives must involve remote access employees or participants, some with only analogue 56k modem access. The idea is to use the likes of QuickTime Streaming Server, which can stream simultaneously to audiences with different access speeds, by sending the rightly-scaled video footage to them. For example, someone with only 56k modem access will receive a broadcast stream that is small, while someone with broadband or Ethernet access will receive the quarter to full-screen footage.

Even if you do run some of your broadcast streams in-house on your Ethernet network, you may want to reduce the video down appropriately so that it does not cannibalize on other users on your network running their daily network tasks such as ftp, Web browsing and so on.

Most media streaming servers such as QuickTime Streaming Server offers error correction so that there is always some degree of buffer stored before the content is streamed to downstream users. This will provide a simple "skip protection" mechanism for viewers. Because of the buffer, it may compensate for the amount of network traffic at the time of the multicast. But do note that no amount of "skip protection" or buffering can compare to a unicast on a dedicated node-to-node connection.

If you have a dedicated leased line, but running between 64 to 128 kbps, you may want to streamline your multicast footage to be compatible with analogue 56k modem rates, since 64 kbps leased lines can be saturated in no time, and you will bring your own connectivity to a crawl.

For a successful multicast to hundreds of users, you may want to consider leasing a full T1 line, or at least dedicate a full-time SDSL for this multicast purpose. Otherwise, you may have to rent multicasting services from the likes of Akamai (www.akamai.com).

Multicasting for e-learning or corporate communications can be very engaging and even addictive. As usual, invest small first, before progressing to studio environments with more professional recordings and broadcasts. It will try your patience, as well as your budget, but the rewards are often worth the effort.

Seamus Phan is research director at KnowledgeLabs News Center (www.knowledgelabs.net), an independent technology news bureau and writes for Network Computing-The Asian Edition. He can be reached at seamus@knowledgelabs.net