Content area
Full Text
SAN JOSE, CALIF. - PC-sub Systems vendors are hailing H.323, the International Telecommunications Union standard for videoconferencing over LANs, as the long-sought grail that will enable rapid deployment of desktop videoconferencing within corporations. But while H.323 ensures interoperability among videoconferencing systems over a network, it can make no similar guarantees for quality of service.
The industry, in the view of some observers, may thus be forced to embark on a new quest: for standards that will allow digital video and audio to be "packetized" in a cohesive way that will assure their delivery in unbroken, readily intelligible streams.
Endorsement of the H.323 specification has been universal among conferencing-chip and -system vendors, and Microsoft Corp.'s recently revised NetMeeting conferencing software is H.323-compatible. H.323 implementations were the order of the day here at the recent Desktop Video Conferencing (DVC) conference and exhibition.
"H.323 will become the lingua france for videoconferencing on LANs," said Andrew Davis, a principal at The Wainhouse Consulting Group (Brookline, Mass. ) and author of the study "Teleconferencing Markets and Strategies," marketed by Forward Concepts (Tempe, Ariz.).
Davis noted that H.323 uses Internet protocols (IPs) to ship video and audio packets on Ethernet networks. Since IP packets are also used as the basic exchange medium on the Internet, H.323 will enable the Internet videophone.
Many companies are converting existing room-to-room conferencing systems from H.320 to H. 323, Davis said. That will enable corporate executives not only to videoconference with their staffs over LANs but also to use the lessexpensive Internet to make crosscountry and overseas calls. Internet access, Davis said, is the fuel driving the H.323 "juggernaut."
But there are no assurances that the digitized and compressed audio/video packets propelled by that juggernaut will be received and decoded intelligibly. Ethernet networks using CSMA/CD, in particular, accommodate collisions among multiple talkers on a network cable. When collisions occur, data transmitters back off and retry at random intervals. But that sets the scene for data-stream disruptions.
The problem is less severe for video than for audio: Many video frames can be dropped before onscreen motion appears "jerky" or disconnected. In the worst case, a stalled video-frame sequence would simply "freeze" an otherwise acceptable video frame on the screen. But speech becomes unintelligible unless it is delivered...