Video compression in transition

Broadcasters and service providers have a wide variety of options to consider when choosing the correct compression strategy for the contribution and distribution of broadcast content.

With significant investment in MPEG-2, broadcasters are keen to see vendors addressing the need for greater SD compression efficiency for distribution applications. Higher channel densities and the trend for larger flat screens have resulted in renewed interest in re-evaluating MPEG-2 for further compression gains in an environment where bandwidth is both scarce and expensive. HD distribution started with MPEG-2 over cable and satellite, but many HD services are now possible as a result of the compression efficiencies attainable with MPEG-4 AVC.

Current compression systems now demand multistreaming and file-based capability alongside real-time delivery. This article will investigate the issues surrounding the selection of a compression standard, optimum operational bit rate and ideal system configuration.

MPEG-4 compression

When first introduced, the MPEG-4 compression standard was heralded by many as an MPEG-2 replacement. Since then, a series of profiles and levels has been developed within the MPEG-4 standard that successfully addresses many of the shortcomings of MPEG-2. These include greater compression ratios to facilitate HD carriage, resilience to errors introduced by packet-based IP distribution, and the application of MPEG techniques for mobile, handheld and PC streaming applications.

MPEG-4 has been refined to focus on commercial areas with requirements that eclipse the capabilities of MPEG-2. The result is MPEG-4 Part 10, more commonly known as H.264 or Advanced Video Coding (AVC). In terms of compression, H.264 is now viewed as a possible successor to MPEG-2 in broadcast distribution applications. However, it should not be assumed that applications will switch from MPEG-2 just to track the latest in compression technology. Many compression applications will remain MPEG-2 for the foreseeable future because of the large installed base of set-top boxes. Additionally, for some broadcasters, the gain associated with upgrading the compression scheme does not yield the benefit to offset the downside of upgrading in terms of capital outlay and retraining.

MPEG-4 AVC was initially developed for low-bit rate multimedia applications. This made the standard more robust and appropriate for IP-based networks when compared with predecessor legacy standards like MPEG-2. This advantage carries through to the application of AVC in contribution applications, and consequently makes bandwidth more plentiful and workflow shorter.

Beyond MPEG-2 compression

MPEG-2 compression is a tough act to follow, so it should not be assumed that broadcast applications will no longer use this standard. Much of the research effort in video compression has been directed toward improving MPEG-2 to exploit the advantages brought about by more capable processing technology.

Where MPEG-2 has met significant challenge is in bandwidth-hungry HD contribution applications, where the savings offered by a move to a more efficient scheme are seen as commercially essential.

In terms of compression, the following are major factors:

  • A significant move toward HD programming;
  • Demand for greater compression efficiency;
  • Greater use of file-based production techniques;
  • Availability of more transmission bandwidth;
  • A desire for reduced workflow; and
  • Support for multiple playout platforms.

These factors make the selection of a suitable compression scheme for contribution far more complex than when MPEG-2 was introduced into this environment.

The bit rates used in a typical production chain are shown in Figure 1 on page 42. The rates shown are typical of those used in HD 1080i/720p environments with distribution over MPEG-2. The main development in terms of bit rate has come about because of a desire to greatly reduce the emission bandwidth using H.264 to rates of approximately 4Mb/s-12Mb/s.

In terms of mastering quality, 1080p acquisition introduces the prospect of 3Gb/s for full frame-rate carriage. So even though there is more capability to carry higher payloads, for the vast majority of applications, there is still a requirement for efficient video compression to bring the bit rate down to manageable levels. So, what constitutes a manageable bit rate in a contribution environment, and what compression standard can be used?

Both MPEG-4 AVC and MPEG-2 have been applied to multiple markets in the typical production chain. (See Figure 2.) This usage in multiple markets has significant workflow benefits in reducing the complexity of each production stage.

In most areas, direct-to-home (DTH) deployments have been based on MPEG-2, creating an established infrastructure and a significant legacy for which replacement of MPEG-2 by AVC equipment is not a practical solution. Broadcasters and service providers need to work further within the constraints of MPEG-2 on infrastructure that needs replacement or upgrade to a more efficient implementation.

Many early DTH systems now require upgrading simply because they have reached the end of their life cycle. Replacement of these early MPEG-2 systems has led many compression vendors to re-evaluate their stand on video quality and to retract their view that MPEG-2 had no further developments in compression efficiency to offer. Traditionally, four strains of MPEG compression systems have been marketed as broadcast products catering to the SD and HD variants for both MPEG-2 and AVC applications. The benefits of applying techniques learned from AVC, along with the extra processing power available, have yielded a new breed of encoders that not only offer enhanced video quality, but support both MPEG-2 and AVC in dense multichannel architectures. The versatility inherent in such a platform is crucial when addressing a market with requirements that center around legacy system support, the aspirations a provider has for distributing content and a continued desire for greater bit rate efficiency in less rack space.

For contribution and primary distribution applications, where content is exchanged between broadcasters or fed from remote events, similar ties exist to MPEG-2. Professional profiles and levels were developed to extend the range of applications for MPEG-2. The most notable extensions involved enhanced chrominance support through the introduction of 4:2:2, carriage of ancillary data and allowance for milder compression ratios to preserve image quality. In these professional applications, the inertia behind the continued use of MPEG-2 remains strong.

AVC is making inroads in these applications, where there is a strong overlap in terms of feature sets between the professional and final distribution markets. This overlap exists in newsgathering where 4:2:0 chrominance sampling and high compression ratios are in demand for carriage over narrowband satellite, terrestrial and IP links.

The appeal of AVC has led many second-generation Digital Terrestrial Transmission (DTT) platforms to evolve from exclusive MPEG-2 SD transmissions. Through next-generation encoders, this evolutionary path adds HD AVC services alongside MPEG-2 and eventually supports an all-AVC approach that embraces SD and HD alongside a wide variety of streamed applications. This approach is currently under way for DTT within the UK and will no doubt be repeated in other regions where MPEG-2 is currently the dominant compression standard.

The future for broadcast headends

Greenfield broadcasters are in the fortunate position of not having strong legacy ties to MPEG-2; thus, they are able to deploy systems from the ground up that are fully based on AVC. The resulting streamlined systems, based on the latest generation of AVC encoder products, has tremendous advantages, including the ability to offer the greatest bit rate efficiency, the highest channel density with good redundancy provisioning and the ability to address multiple markets by providing simultaneous streams for a wide variety of platforms.

Video quality is retained within systems that allow content to remain within a particular compression standard. Broadcast content can then be manipulated by adjusting parameters such as frame rate, aspect ratio and compressed bit rate to make them appropriate for the target platform. While this approach is commonly used for broadcast MPEG-2 content, AVC-based systems become very attractive due to the reduction in workflow steps, streamlined infrastructure and retention of picture quality.

Regions that do not have a large MPEG-2 legacy infrastructure can deploy AVC directly with obvious benefits. However, with the vast majority of broadcast content existing as MPEG-2, digital turnaround of such material is a common requirement. Decoding MPEG-2 content has often been performed by separate decoders, but increasingly this functionality is integrated within the encoder. High-end encoders not only have the capability to offer significant coding gains for multiple channels, but also they can now deal with a wide variety of input and output formats. (See Figure 3 on page 46.)

This flexibility is further extended by offering dual encoding channels for each video input. Normally, a low-resolution path has been optionally offered alongside the main video encode chain to allow picture-in-picture capability. Making the second channel a fully featured encoder and adding an up/downconverter to the input of the second channel allows a single encoder to produce two encoded outputs. This functionality is commonly requested by broadcasters and service providers. (See Figure 4.)

Rather than viewing AVC as a challenge to MPEG-2, companies are adding functionality to encoders to ease the introduction of more AVC-based services alongside MPEG-2. In the long term, MPEG-2 infrastructure might well be replaced by AVC, but for the moment, the two standards will coexist.

Ian Trow is director of broadcast solutions at Harmonic.