Internet2 Virtual Briefings

powered by the Internet2 Commons


Lessons Learned

• The Internet2 Virtual Briefings serve several purposes:
• Provide a testbed for using and debugging new collaboration technologies
• Demonstrate and promote the widespread use of new collaboration technologies
• Provide a platform to deliver interactive briefings of Internet2-related activities

Our objective for the Internet2 Virtual Briefings is to find the optimal mix of quality, audience reach, and interactivity with the technologies we use to provide the Briefings. We also want to reach a broad range of participants based on the content of the briefings. These goals mean that we must use a mix of technologies to support these events, rather than just one solution. Because these Briefings are experimental in nature, we feel that the Internet2 member community might benefit by sharing in some of our "Lessons Learned."

Ongoing Challenges
Virtual Briefings are currently H.323 MCU based and utilize The Internet2 Commons videoconferencing environment. The Briefings are netcast in a wide variety of formats, including MPEG-1, MPEG-2, Mbone tools, raw digital video (DVTS), RealPlayer streams, Vbrick, etc. Some limitations we've experienced are lack of:

• Titling
• Higher quality video (ie, MPEG-2 or better quality)
• Automatic supression of noise
• Interoperability among a variety of technologies
• Tight control of production – who gets to speak, what is displayed
• Floor control mechanisms
• Text/slide presentations (e.g., PowerPoint)
• Easy to use scheduling
• Multicast at some locations
• Resilency to packet loss
• Remote control (i.e., web control)
• Reliable and interoperable H.323 equipment

Internet2 continues to research solutions to these limitations. One of our end goals is to overcome these limitations and provide an easy-to-use videoconferencing and netcasting system.

Technical Goals for October 2002 Virtual Briefing
For the October 2002 Virtual Briefing, our technical goals are to:

• Continue the current level of functionality
• In addition to the netcast of the Briefing presenters, provide a second netcast showing the presenters' slides (most likely available through Windows MediaPlayer and vic).
  

Challenges We've Faced at Previous Virtual Briefings
During previous virtual briefings we encountered problems including the following:

• Connectivity failure at Internet2 Ann Arbor office (changes to router, insufficient testing).
• Real streaming server poor performance (installation of firewall, insufficient testing).
• Connectivity failure at Internet2 Armonk office (power outage).
• Open mic noise and inability to identify source (inadequate MCU features).
• Packet loss (overloaded links, duplex mismatch).
• H.323 quality (largely solved by requiring the use of 768K and H.261).
• Lack of RTP/TV box robustness and problems with A/V synchronization.
• Poor lighting, camera position, audio quality (lack of prior testing and time).
• Audio feedback (user error).
• Names not visible in RAT (Polycom H.323 protocol error) handled poorly by client.
• Network flooding (Ethernet switch doesn't handle multicast traffic with no listener).
• Power failure (lack of UPS).
• Freezing video in MCU-based calls (MCU software or client problem).
• No audio (Via Video problem or modifications to PC, lack of ability to test "next speaker").
• H.323 Gateway allows an alias to have the same name as the conference ID, resulting in confusion for the gatekeeer.
• Multicast is often down, although the lack of a beacon listing doesn't indicate this.
• Network endpoints are needed at both ends of the network for testing.
• Video freezes be switching to another speaker locations (MCU problem).

Technical Debriefs from Previous Virtual Briefings

13 September 2002 Virtual Briefing — During our September 2002 Virtual Briefing we continued the current level of functionality from previous briefings and experimented with local video switching and "mix minus" audio. The biggest challenge with this Briefing was that we were unable to receive video from the Georgetown University presenter. Therefore, for future Briefings, the production staff recommends:

• Perform testing using the same presenter room and MCU that will be used for the actual Briefing.
• Briefing production staff should reboot all equipment one hour before the Briefing starts.
• Presenters should reserve their rooms for at least one hour before the Briefing starts, to allow for testing and trouble-shooting, and for 30 minutes after the Briefing is over, to accommodate any overflow discussion and also for an immediate debrief with Briefing production staff.
• Briefing production staff should have slides and pictures of presenters available to display locally, in case of video or audio transmission problems.

Additional Lessons Learned from the 13 September 2002 Briefing are available here.

17 July 2002 Virtual Briefing — During our July 2002 Virtual Briefing we continued the current level of functionality from previous briefings and added MPEG-4 unicast streams using Microsoft's streaming server. The challenges we faced during this briefing included confusion resulting from the H.323 Gateway, because it allows an alias to have the same name as the conference ID (which confuses the gatekeeper). Multicast was often down, although the lack of a beacon listing didn't indicate this. Video occasionally froze before switching to other locations, this was attributed to an MCU problem. For future briefings, network testing endpoints are needed at both ends to ensure adequate performance.
Technology Diagram from the 17 July 2002 Virtual Briefing [pdf] [jpg]

11 June 2002 Virtual Briefing — During our June 2002 Virtual Briefing we experienced problems with audio feedback (due to user error) and freezing video in MCU-based calls (due to MCU software or a client problem).
Technology Diagram from the 11 June 2002 Virtual Briefing [pdf] [jpg]

Here is a complete description of the technology strategy for the Internet2 Virtual Briefings