Latest News

Compression Challenges in the Broadcast Environment 2.0

By George Lawton | April 1, 2013

      There are a number of trends in the broadcast environment relating to the evolution of the codecs for compression and managing the workflow to support more channels and devices. At the high end, the publicity about 4K promises to require considerable new bandwidth for broadcasters, prompting the need for better compression tools. There are also some efforts being made to improve quality in the interim with better sound and higher frame rates.

      MPEG-4 promises better compression than MPEG-2 and is gaining considerable traction by replacing the latter in most broadcast environments. Major broadcasters, such as DirecTV, have put considerable resources into the transition to MPEG-4. Romulo Pontual, CTO at DirecTV, says they began broadcasting in MPEG-2 in 1994 and introduced HD in MPEG-4 about ten years later. Today, all of DirecTV HD broadcasts are delivered in MPEG-4 and most of the SD content is still transmitted using MPEG-2, although some local content is also broadcast in MPEG-4.

       

      H.264 (MPEG-4) Getting Traction

      “People are making plans or starting to adopt MPEG-4 everywhere,” says Benoit Fouchard, chief strategy officer at Ateme, a video compression solutions provider. “Broadcasters have so much to gain from the cost of using satellites that whether they are broadcasting HD or SD, they have adopted MPEG-4. In cable, the operators are weighed down by the number of boxes deployed that only support MPEG-2.”

      Broadcasters can sometimes cut their satellite communications requirements in half by moving to MPEG-4 for their satellite feeds. But Fouchard adds that a small operator with fewer channels might be more cautious.

      There is also ongoing work to squeeze even more compression out of MPEG-4 using AI. Rodolfo Vargas, CEO at eyeIO says, “We have been able to demonstrate 50 percent more compression than what could be achieved using traditional techniques. Current compression approaches are very mechanical and based on techniques to compress using macro blocks. However, now we can use AI techniques such as bit-level segmentation and computer vision techniques to understand what is on the image.”

       

      HEVC (H.265) Shows Promise

      HEVC is the promising successor to H.264 and is expected to allow a 30 percent improvement in compression performance. As broadcasters move into higher resolutions with 4K TV, HEVC promises even better gains, perhaps as much as a 40 percent improvement over MPEG-4, according to Fouchard.

      But HEVC is not standardized and will not be available in industrial grade equipment for a number of years. The types of implementations of HEVC available today are more for the lab or research purposes. “Most people only throw in their best effort once the standard has been stamped,” notes Fouchard.

      The HEVC standard is expected to go through two stages of implementation. The main profile will be standardized first followed by a high profile version of HEVC in a few more years.

      Some of the attraction of HEVC’s higher performance may also be muted by improvements in H.264, which is compatible with existing decoders, says Vargas. “Although it has been reported that HEVC allows a 50 percent better compression efficiency than H.264, we are only getting about an 18 percent improvement using the same profiles.”

      PBS has just gone through the transition from MPEG-2 to MPEG-4 for distributing public TV over satellite to local broadcasters in the United States. “We were due for a capital refresh and saw the viability of the technology,” John McCoskey, CTO of PBS says.

      He also notes that there were two aspects of the technology that were attractive to PBS. First they could increase the quality of their services by allowing both subjectively and objectively higher quality for stations. And second, they were able to reduce the satellite capacity to provide these higher quality services. “We went from a distribution that used 4 Ku-band transponders to one that now uses 3 Ku-band transponders while at the same time increasing the quality of HD and SD channels and adding new HD service we did not have before. This was a notable change in our monthly recurring bill,” he says.

      Traditional broadcasters like to take a long-term view when upgrading their compression. “When I look at equipment like this lasting us five years on the capital side, it is almost an equal share between capital cost and the cost of satellite capacity,” says McCoskey.

      PBS was able to make the transition work out smoothly by planning for it. For several months they kept both services up and running so it was not a light switch change; each local station could make the transition at any time in that window. There were a few stragglers, but they we were able to get the entire system transitioned to MPEG-4 during this process.

      During the transition, one additional transponder was secured to provide the minimum service required across both MPEG-2 and MPEG-4 streams. The primary PBS channel was duplicated, but channels doing soft feeds, such as promotions were not.

      In the PBS system, the HD MPEG-2 feed was transmitted at 17 Mbps for video and 324 Kbps for audio. With the transition to MPEG-4, the data rate was reduced to 12 Mbps for video, and the audio was boosted to 448 Kbps.

      “Although that is significantly smaller than the 17 Mbps, the quality improvement was remarkable. That was one of the goals to boost quality of the service that would be distributed out to broadcast towers,” McCoskey says.

      If they had chosen to keep the quality the same, they would have been able to transmit MPEG-4 at half the rate of MPEG-2. For SD, PBS adopted two rates, depending on the services of either 4.5 Mbps or 3 Mbps, which were reduced to 2 Mbps. The ATSC HD stream is broadcast at 19.4 Mbps using MPEG-2 over the airwaves. The PBS stations transmit at 10-17 Mbps for video. The different data rates vary by whether they are using 1080i, 720p, and on how many multicast stations they are running.

       

      ATSC Evolution

      Broadcasters are also keeping an eye on the evolution of ATSC standards, which promise higher quality video with better support for interactivity.

      “We are interested in the ATSC 2.0 efforts, which are backwards compatible with the current standards,” says McCoskey. “Going forward, the ATSC 3.0 standard will enable technology that does not have to be backwards compatible with ATSC 1.0.”

      ATSC has an option to bring in H.264 support for encoding higher quality or for reducing capacity requirements. It also brings in elements around non-real time distribution and supports too for feeding second screen applications running on a tablet or smartphone. “The key with 2.0 is that it is backwards compatible. If you put up a signal meeting the 2.0 spec, it will not break receivers that are 2.0 compliant,” he says.

      Going forward, he expects their next capital refresh to be based on HEVC. “We are seeing more implementation of the base technology. There is now a lot of silicon exploiting it,” McCoskey says. “For us, it is a natural evolution, just like the transition from MPEG-2 to MPEG-4. I think we will see HEVC in the next two years at the most. When you think about the age of MPEG-4 and when people first transitioned, we will see people transition to HEVC in a 2-year time frame.”

       

      Hype

      Depending on whom you ask, 4K systems will require about four to eight times the bandwidth that HD does today. One of the early challenges in the move to 4K has been confusion in its meaning, according to Vargas. “Some camera companies are announcing doing 4K, but might define a pixel as representing only one of the primary colors. There is some confusion in the market as some are calling this 2K, while others call it 4K. It is similar to the way in which people use the term HD to describe 720p rather than 1080p. We don’t want that same kind of confusion with Ultra HD,” he says.

      More than half a million 4K×2K TVs are forecast to ship worldwide this year, growing to more than 7 million by 2016, according to the latest Quarterly TV Design and Features Report from NPD DisplaySearch. Sony, LG, Samsung, and Sharp are all moving forward with this technology.

      As a result, Paul Gray, director of TV Electronics Research for NPD DisplaySearch says that “investments are beginning to ensure that 4K×2K content is readily available to consumers. Several announcements about proprietary 4K×2K streaming and download services were made at CES, however, satellite and cable services from established providers will take some time.”

      Eutelsat Communications recently launched the first dedicated Ultra HD channel across Europe on their Eutelsat 10A satellite. This new channel operates in progressive mode at a frame rate of 50 frames per second. All the content is encoded in MPEG-4 with the help of Ateme and is transmitted at a bit rate of 40 Mbps in four Quad HD streams. The transmissions are uplinked from Eutelsat 10A’s teleport located at Rambouillet, near Paris, France.

      There is some interest in taking advantage of not only 4K larger screen size, but also support for better color, according to Vargas. He says that operators want to move from 8-bits to 10-bit pixel encoding, which provides 25 percent more dynamic range. There is also interest in doubling the chroma from 4:2:0 to 4:2:2. “This lets you see the video with greater quality such as the subtle texture of hair,” he says.

       

      Frame Rate

      As the full rollout toward 4K could take a while, many in the industry are looking at other interim ways of increasing quality to attract customers and differentiate their video. Some broadcasters may choose to use the extra efficiency of better codecs to transmit 60 frames per second, as opposed to the 30 fps commonly delivered today for high-motion coverage like sports.

      “Our research shows that a step improvement in user experience would come sooner from higher frame rate than high resolution pictures,” Fouchard says. “The quality of viewing sports has more to do with making a step in terms of frame rate than resolution.”

      Depending on the motion in the picture, a wide-angle view might require only 30 percent more bandwidth for double the frame rate, whereas high motion tight-angle scenes might require 70 percent more bandwidth.

      Fouchard argues that most cameras and HD TV sets support these higher frame rates today, but there are challenges with the decoders. Some set-top boxes will only need a firmware update, while others will never support it and will need to be replaced. MPEG-4 by design is intended to support a higher frame rate at a fraction of the incremental bandwidth.

      At the high end, 120 Hz is available in TV sets, but is not widely available in decoding chips, so Fouchard does not expect to see it in the near future. “Even though we can perceive differences going up to 240 Hz frame rates, I don’t expect to see this enabled by the industry. It would be too hard a sell to incorporate into set-tops right now. For now, doubling or quadrupling frame rate would be great and would make everyone happy,” he notes. Fouchard expects to see 60 Hz transmission to be available between professionals soon and believes that it would be relatively easy for an operator to source high frame rate content from a broadcaster like ESPN or experimental content like a Word Cup game with little technical difficulty.