Scientific Visualization
VideoOverIP technology enables collaboration between remote users
By John Vines, Audiovisual Production Specialist
The ARL MSRC has strong ties to the HPC user base and a history of being at the forefront of high performance computing. Presently, the center incorporates state-of-the-art, massively parallel IBM, SGI and Linux computational assets and leading-edge networking.
OK, so maybe you already knew this, but I bet you don’t know that the MSRC not only helps solves the DoD’s most complex computational problems but also helps solve another problem: finding solutions that provide real-time communication and collaboration among geographically dispersed personnel.
VideoOverIP Technology
In the fall 2002 Link [“ARL MSRC introduces MediaNet services,” pg.7], an article discussed how the MSRC was leveraging technology that incorporates IP, internet protocol, and communication technologies, specifically video over Internet Protocol (VideoOverIP).
The VideoOverIP technology is being tested and developed on a regular basis to connect two of the ARL MSRC's reality centers, one at the Aberdeen Proving Ground and the other one at the Adelphi laboratory site, and with remote users in California. Through the use of this technology, we have been able to interact with personnel at the remote location in real-time.
VideoOverIP is also used nearly exclusively to collaborate with remote users on projects, including scientific visualization, video, and DVD productions. The ability to send a signal in real-time from the graphics of our primary SGI visualization machine has increased our capabilities and efficiency. In the past, we've dumped images and sent MPEG files for researchers to review; the process may take 4, 5 or 6 iterations to get the data visualized in a meaningful manner. The VideoOverIP technology has enabled us to visualize the data in a meaningful manner the first time.
With the leading-edge network connectivity at the MSRC–not to mention a second-to-none network team–IP communications is an exciting area.
Examples of networked compressed video and audio are video conferencing, streaming video, videophones, video on demand (VOD) and distance learning. You may wonder, why compress? Well, without the use of compression, low-quality color videophone applications with 352-by-288 pixels resolution at 15 frames per second would require approximately 18 megabits per second. Digital-television applications, video teleconferencing and remote collaboration sessions, at standard definition resolutions 640-by-480 pixels at 30 frames per second, would require as much as 100 megabits per second. If we were to stream uncompressed high-definition (HD) signals, we would use 1.5 gigabits per second. Not that our network couldn’t handle it but compression would help.
So we know how much uncompressed signals require, but what type of compression are we using and how much bandwidth will it save? To start with, we will be using MPEG compression, specifically MPEG 1, MPEG 2 and MPEG 4. MPEG 1 provides resolution of approximately 352-by-240 pixels at 30 frames per second at 1.5 megabits per second and a VHS-quality picture. MPEG 2 provides resolution of approximately 640-by-480 pixels at 30 frames per second at 4 megabits per second. (You’ll read the word “approximately” when it comes to encoding MPEG streams because MPEG supports many sizes and formats. The explanation of which would take the entire article, so you’ll have to trust me.)
The MPEG 2 format can provide better resolution and higher quality at higher bit rates. Our default MPEG 2 stream is encoded at a resolution of 640-by-480 pixels at 30 frames per second at 4 megabits per second. The video's quality is similar to a DVD's, with stereo audio encoded at 44.1 or 48 kHz.
Now that we know more about the compression, quality and bandwidth requirements, let’s discuss setting up a session.
First, to establish an IP link between equipment and systems, the users on both ends must be able to see each other; encoders must be able to talk to decoders, and decoders must be able to talk to encoders. If a remote user is using a PC client, then the PC must be reachable from the MPEG server and the server’s IP address must be reachable from the PC. Obviously the network team is worth its weight in gold when it comes to setting up and configuring these systems.
The biggest issue is the bandwidth for requirements for remote users; it’s obvious that your connection is only as good as your smallest network connection. There needs to be some testing between sites. Even if you have a map of the entire network with up-to-date bandwidth information, you will still end up testing and tweaking stream and network paths to ensure the best possible signal with the lowest possible latency.
Defining HDOverIP
VideoOverIP also operates at HD resolutions and are referred to as HDOverIP systems. The setup for these systems is much the same as for the VideoOverIP, except you are using HD equipment and signals.
HDOverIP system has advantages and disadvantages. The advantages are increased resolution, clearer images and, overall, a much higher quality signal. Not surprisingly, HDOverIP technology lends itself to a computer-graphics environment. The ability to output a 720p, 1280-by-720 pixels, progressive-scanned image from graphics obviously provides a much higher resolution image.
Associated costs are much higher: approximately $80,000 for an HD encoding and streaming station and approximately $1,200 per PC client. Not to mention that the network bandwidth requirements change from 4 to 6 Mbps to approximately 19 to 24 Mbps.
'Bursting' on the Scene
The latest development in the use of IP communications is from a company called Teraburst Networks. Teraburst has been at the forefront of optically transmitted computer graphics video for many years and has managed to compress video signals at computer-graphics resolutions and provide this signal over an IP network.
Teraburst is reporting the capability of encoding computer graphics video, keyboard and mouse events into a real-time MPEG 4 stream. Specifically, the V2D equipment is reported to handle resolutions of 1600-by-1200 pixels at up to 110 kHz horizontal frequency. The ARL MSRC has ordered two V2Ds and is planning to install them in the Reality Center here. So more to come on this technology later.