Figure 1. A typical master control system today. Note the presence of numerous encoders and decoders in an environment. Click here to see an enlarged diagram.
Compressed digital video used to be considered a novelty in the television industry. But numerous technology advances, not to mention the FCC mandate for digital terrestrial service and increasing consumer demand for high-quality images, have made it a must-have for broadcasters.
Processing compressed video, however, still presents a number of challenges. This is why it is typically decoded to baseband, processed and then re-encoded. In fact, by the time it is ready to air, some footage has been encoded and decoded so many times that the original quality has been significantly degraded. This is an expensive solution, especially in a high-definition or multichannel environment.
What, then, are broadcasters to do when confronted with multiple video sources all compressed to different bit rates, particularly if they wish to place their own stamp on this programming without further degrading the quality with yet another decode/re-encode cycle? Is there a way in which revenue-generating services, such as ad-insertion, interactive metadata and graphical overlays, can be added to compressed bit streams at master control?
Finally, the answer is yes. A number of new technologies developed over the past several years have come together to provide seamless integration of video and audio processing elements into compressed digital bit streams. These new technologies solve the problem of switching between compressed feeds, in various formats and bit rates, which would otherwise overload the decoder buffer, causing picture freezes, glitches and audio popping. The result of combining these technologies is the Terayon BP-5100 broadcast platform.
One of the key technologies is statistical re-multiplexing, which requires expertise in rateshaping. This technology grew out of the cable industry's need to take existing compressed satellite feeds and recombine them to serve different multiplexes on their networks. Rateshaping is the ability to intelligently change the bit rate of the compressed feed without noticeably altering the picture quality. A 4Mb/s feed could be scaled down to 3.5Mb/s without much difficulty, and the extra bandwidth could then be devoted to another bit stream that may require more data at any given moment. But more significantly, rateshaping can be used to maintain the MPEG reference decoder model to ensure that either an underflow or overflow to the decoder buffer does not occur. This is a crucial function if broadcasters wish to alter incoming feeds at master control in the compressed domain.
Figure 2. A compressed digital master control system affords a more streamlined approach to signal management. Click here to see an enlarged diagram.
Several other technologies also play a part in compressed domain master control. The ability to switch between MPEG streams plays a crucial role through the use of advanced predictive techniques that identify the correct I-frames to allow the new stream to be inserted in a single cut. This avoids the blockiness, latency and black frames that occur when cuts are attempted to a B- or P-frame.
Also vital was the development of a new keying technique based on selective decoding/re-encoding. Only a small portion of a frame is decompressed to add a graphic or logo without degrading overall picture quality.
How does all of this work in practice? Quite smoothly. Let's take the example of digital program insertion. Because we can now perform this operation on the elementary stream level for both audio and video, we can insert interstitial video seamlessly in the compressed domain. The SCTE 30 digital program standards specify the interface between the video server and the splicing device, while the SCTE 35 digital cue messages allow precise control over where we insert video and for how long.
As long as the MPEG group of pictures sequence is conditioned properly, we can offer frame-accurate insertion. It is now possible to take an arbitrary slice of compressed video and seamlessly insert it into another slice with no visible artifacts.
There is little reason for broadcasters not to embrace compressed domain master control switching. It streamlines the on-air process by greatly reducing the number of encoders and decoders in a typical master control room, and it preserves picture quality by cutting the amount of processing required to get the signal to the home. Add in the facts that it promotes greater localization of network feeds and provides for smoother ad-insertion to boost local revenue, and the technology is a clear winner.
Michael Adams is the vice president of video architecture and technology at Terayon Communications Systems.