SMPTE Showcases Video Over IP

February 21, 2007

(click thumbnail)
At the SMPTE/Vidtrans Interop, monitors show impairments and how they were corrected to demonstrate how manufacturers handle forward error correction.
IP has emerged as a mainstream application for professional video, as evidenced by the papers and demonstrations at last month's annual SMPTE Digital Content Connection & VSF VidTrans Joint Conference.

In addition to new standards, video over IP technology was displayed and discussed at most of the booths and the subject of many of the papers. These discussions are no longer theoretical--broadcasters are starting to deploy video over IP for contribution, distribution and direct to consumer applications.

"Anyone who was skeptical about broadcast-quality video transport over IP had to come away from this conference with the realization that this technology has been embraced and is ready to go to air," said Paul Atwell, director of video networking for Ion Media Networks, formerly Pax.


Broadcasters are coming to recognize that many different things can happen to their pristine video signals when they are sent over an IP network.

In response to these concerns, the conference hosted a COP3 interoperability test, which took place on the exhibit floor.

The (now-inactive) Pro-MPEG forum developed the standard that is known as Code of Practice 3 Release 2, or COP3-R2. This work has now been taken over by the Video Services Forum (VSF) and SMPTE, and designated SMPTE 2022.

In the project, an independent jury evaluated the ability of each participating exhibitor's encoder to create a stream that could be correctly decoded by every other vendor involved in the interops. Controlled errors were injected in the interconnecting network to exercise the error correction capability of the added forward error correction (FEC). In all cases judged by an independent jury, including this reporter, all of the injected errors were removed and a clean video signal was produced. The five vendors (Harris, Grass Valley, Tandberg Television, Path One and T-VIPS) supplied FEC encoders and decoders based on COP3-R2.

A total of 20 tests were performed using video source material supplied by Fox. Each vendor took the compressed video stream used in the test and ran it through its COP3-R2 encoder, which added FEC packets to the video stream. The outputs were then sent to a network impairment device that simulated a network that had a high enough error rate to completely test the error correction limits of COP3-R2. The impaired streams were then delivered to a decoder in each booth, which was responsible for using the FEC data to completely correct the compressed video stream, and then delivered to the display stand for viewing.

"The success of this interop clearly demonstrates the suitability of the COP-3 R2 standard for sending high-quality video streams over an IP network experiencing a high error rate, and the momentum within the broadcast community for supplying this equipment," said John Dale, senior product line manager for Milford, Conn.-based Media Links, a subsidiary of Japanese digital broadcast equipment supplier Media Global Links, and co-chair of the ad-hoc group that hosted the testing.

The interop demonstrated the ability of the COP3 algorithm to operate in the presence of an abnormally large amount of network errors. It also demonstrated that the specification could be understood and followed by a number of competing vendors; no mean feat considering the number of different aspects of the signals that needed to be specified.

In addition to the tests, a number of papers presented at the conference discussed forward error correction. Thomson and Digital Fountain explained how adding a small amount of FEC data to a video stream can allow the receiver to completely recover from a number of errors ranging from simple bit errors to entire lost packets. This process can be tuned, allowing more FEC packets to be used when error rates are high, and less packets when the error rates are low. With the ease and flexibility of FEC, it's hard to imagine why broadcasters would even consider sending high-quality video over an IP network without some type of fairly serious error correction to protect their signals.


A panel discussion led by Brad Gilmer, president of technology consulting firm Gilmer & Associates, explored the conflicts between content owners and content consumers. Some panelists took a hard line, saying that consumers shouldn't think that they own the music on any of the CDs that they bought--they merely own the rights to listen to the performance contained in that CD, which may not include the right to copy the CD or transform it into another form, such as an MP3 file. Others, while not quite so emphatic, were also concerned about protecting the rights of content owners and making sure that consumers don't have the temptation to violate copyrights.

Watermarking, the practice of inserting hidden data inside a digital video or audio stream to give it a unique identification number, was also discussed. Some content providers view watermarking as a solution to digital piracy, enabling content owners to specifically identify when and where copies of content originated, simplifying the process of prosecuting violators. Unfortunately, as Reza Rassool from Widevine pointed out, the processing load to create a unique watermark for each viewer can be staggering unless properly distributed in the distribution network.


Dr. Michael Isnardi of Sarnoff Corp. in Princeton, N.J., gave a fascinating presentation about the work that he has been doing with AT&T, Thomson, and Alcatel on the Video Enhanced DSL (VeDSL) project funded in part by the U.S. Government. The project is designed to explore and demonstrate technologies that would allow a standard 1.5 Mbps DSL circuit to handle two SD program streams simultaneously. To do this, researchers needed to create some significant advances in technology including:

  • A reduction in bit rates of acceptable quality video to 0.07 bits per pixel, representing a threefold improvement over MPEG-2 and a significant improvement over MPEG-4 AVC;

  • Development of a video-aware DSL Access Multiplexer (DSLAM) that can selectively drop IP packets in a network overload situation to negatively affect the video stream; and;

  • Creation of a fast channel change system that would eliminate the one to two second delays found in many digital satellite and CATV services.

  • In developing fast channel change, researchers discovered that using a low-resolution video image with a short GOP compressed to a couple of hundred kilobits per second resulted in a much quicker channel change scheme. Whenever a user decides to change a channel, this stream is delivered to their set-top box, and upconverted for display on the television. This continues until an I-frame arrives on the standard-definition stream and the signals can be spliced back. This low-resolution stream can also be use for picture-in-picture applications by the consumers' set-top boxes.

    "The Video-Enhanced DSL system offers a compelling suite of compression, congestion management and fast channel change technologies that keep end-user quality of experience high without the added expense of over-provisioning network resources," Isnardi said. "The team of Alcatel, AT&T Labs, Sarnoff Corporation and Thomson Corporate Research look forward to furthering the adoption of these technologies within the IPTV industry."

    Receive regular news and technology updates.
    Sign up for our free newsletter here.