By using internet protocol technology in the broadcasting industry, direct-to-home operators have been given increased head-end architecture flexibility. This added flexibility is not enough when faced with the challenge of having to deliver content over multiple networks to maximise customer reach. Content consumption is increasingly spread across a variety of delivery and play-out platforms requiring flexible network architectures on which to distribute programming. Next generation hybrid DTH networks draw upon the best of breed head-end technology to meet broadcasters' needs to deliver content in this new environment. Only by a tight coupling of IP and the latest advances in compression, content repurposing, bitrate shaping and multiplexing technologies can broadcasters deliver next generation hybrid multi-network head-end architecture.

A changing viewing landscape

Television viewers are changing and with a growing new internet generation. Customers are getting more fragmented and increasingly moving away from linear Television in favor of on-demand content. In addition, viewers are consuming content in an increasingly fragmented way compared to the traditional living room. Content can be viewed on anything from mobile phones to large HD screens. The linear Television model is far from broken but is getting strong competition from streaming and downloads services. IPTV has the prospect of addressing this diverse service offering, but the demands on broadband delivery to meet the channel bandwidth and quality requirements are difficult to meet, especially with legacy infrastructure. IPTV has made significant in-roads to establish market share.

Satellite and cable broadcasters were early to explore new service offerings like VoD and Near Video-on-Demand and have pushed into the broadband market for additional revenue through triple play offering. By establishing a two-way communication link to the consumer, interactive services to meet the changing viewing landscape can be made a reality. With an increasingly diverse customer base content need to deliver in a multi network environment. Next generation hybrid head-end will require more capacity and flexibility to meet today's demand and be able to evolve over time to meet new requirements.

High quality IP based delivery

IP based video delivery and transportation is getting more common due to the increased flexibility it is offering. As any broadcast head-end will need to meet the demand for future expansions through multi-network delivery, next-generation head-end will need the flexibility that IP is offering. Using asynchronous serial interfaces for video transportation between different devices within the head-end is losing its appeal. The trend now is to move towards an IP (packetized) transport system with some obvious benefits: the distance limitation of ASI is removed, increased routing flexibility and content availability and improved reliability through new redundancy scheme.

These improvements were initially driving the migration towards IP based head-end architecture. In recent years with the introduction of new products, the adaptation of IP has been taken one step further. With next generation encoders and transcoders capable of combining MPEG-2 and MPEG-4 compression with an all-IP interface configuration, the head-end becoming truly flexible. Traditional encoder with SDI input and ASI out, will now have IP in and IP out and instead of uncompressed video feed, the content is received pre-compressed. This opens a whole new range of possibilities for head-end architectures, namely IP based distributed head-end architecture and multi-network video compression.

IP based distributed head-end architecture

Statistical techniques have been used to aggregate video content at broadcast head-ends for some time now. Achieving significant reductions in operator costs, to maintain competitive position, is the final consideration facing broadcasters and Internet service providers. These demands have led to statistically multiplexed video systems utilising IP infrastructure, in addition to dedicated broadcast links, to adopt a distributed topology allowing source encoders to reside at remote broadcaster sites, rather than a centralised head-end.

The broadcast industry has used expensive and proprietary hardware to provide statistical multiplexing features in networks since the widespread use of MPEG-2 based systems. With the advent of IP-based networks, the notion of multiplexing compressed sources that are co-located at a broadcast head-end is becoming less important, while the requirement to fit multiple channels in a bandwidth-constrained network allowing for encoders to be remotely located is increasingly important.

Centralised architectures have had the disadvantage of requiring high bandwidth links feeding the head-end prior to statistical aggregation of the content. This results in expensive duplication of compression stages, both at source origination and head-end aggregation points. While the cost of bandwidth has reduced significantly, video is still bandwidth hungry, especially when factoring additional quality during the initial compression stage to counteract the subsequent encoder or decoder loss between contribution and distribution compression stages.

With the next generation head-end, the encoders can either be located at remote sites or a central head-end. The key to being able to decentralise statistical multiplexing systems lies in the synchronisation of control and signal sources across multiple sites.

Multi-Network video compression

Video encoders have traditionally been designed to serve individual networks with little cross purposes. Traditionally, most broadcasters were using standard definition and MPEG-2 compression and the network had minimum changes. With the introduction of MPEG-4, new high definition networks, MobileTV services and IPTV, this was changing. Bandwidth costs money, so any advancement that conserves quality and allow for more channels is key. MPEG-4 has seen widespread adoption in green field pay TV applications where legacy and legislation haven't compelled the use of MPEG-2. A halving of the compressed bit rate along with the use of a new media friendly compression standard, MPEG-4 has proved compelling to those making top level systems decisions.

As encoders capable of providing content in multiple formats are becoming available, so is the possibility to serve multiple networks from a single encoder. The obvious advantage is to reduce the encoder investment but by using this approach, you are also greatly simplifying your head-end architecture. Instead of separate head-end system per network served, these advances allow you to serve all your networks from a single head-end. This however, is not possible unless you move some of the processing to the multiplexer. Most of the new IP centric delivery networks like IPTV and MobileTV have moved away from a multicast environment traditionally used by cable and satellite broadcasters to a Unicast format where only individual channel is stream to each viewer. This means techniques like statistical multiplexing no longer can be used. An encoder operating in a hybrid multi-network environment will need to handle operating both as part of one or multiple statmux pools in addition to serving individual streams to say an IPTV network.

Increasing screen sizes demand higher picture quality

As the screen size is increasing, so is the visibility of video compression artefacts introduced by the encoder. The obvious answer to this trend is to introduce high definition video. High definition has been increasingly made available through BluRay and HD gaming consoles like Sony's PS3. HD is also becoming available as streaming services via the internet. YouTube has started making some of the content available in HD and online Video rental sites like Apple's iTunes has an extensive library of HD films and programs.

High definition does however put large strains on the available transmission bitrate. With some reports estimating that satellite transponder demand by 2015 will outstrip the supply, video compression will be critical to any broadcasters. No delivery mechanism can support 1.5 Gbps demanded by HD signals all the way to the home. Broadcaster's desire to efficiently use bandwidth has driven the development of video compression standards and algorithms. H.264 is the part of the MPEG-4 standard of most interest for broadcast DTH applications offering the prospect of the most advanced compression toolset to deliver quality video in bandwidth constrained environments. DTH is currently dominated by the predecessor MPEG-2 compression standard. MPEG-2 has been extremely successful and is delivering the vast majority of DTH delivered content globally. Replacement of existing compression infrastructure is often viewed as an unacceptable option to broadcasters who have recently finished the installation phase and looking to recoup significant investment in MPEG-2 equipment. High Definition has been the big application area for H.264 and hence the area where a large number of compression manufacturers have concentrated effort to establish the standard. Specifically for HD, H.264 has achieved widespread success in DTH applications. The large benefactors to the improvements offered by H.264 are cable, satellite and IPTV operators wishing to introduce cost effective HD services. H.264's big break over MPEG-2 was that HD services had not been economically viable with predecessor compression standards and so H.264 had a clean run at the market without having to displace widespread MPEG-2 legacy infrastructure.

Using RoI for smaller screen size

A large proportion of internet streaming and mobile phones are now using MPEG-4 for compressing the video. In most cases however re-purposing the content is also needed. Traditionally video is down sampled to fit to the required resolution but this down-sampling process can have some real problems. If we reduce the screen resolution of say a football game too much, you are in danger of making the ball smaller than 1 pixel and hence invisible. Text and graphics would have the same problem.

A new technology solves this issue by using a technique called Region of Interest. RoI is based on model of the human eye to dynamically analyse a picture to form a visual map of the most important part of the picture. This map can be used in a number of applications but in this section describing how it is used within the context of reduced screen size.

By integrating RoI into an encoder, it is possible to form a number of tasks. As the map will give an indication of the most important part of a picture, it is possible to do a more intelligent down-sampling based on this information. Instead of a direct down-sampling, the picture is first cropped to include the most important part of the picture. Then a second stage can form the down-sampling which due to the cropping will maintain the quality and the important part of the picture. This technology can be found in some encoders in the market and form an important part in preparing and optimising the content for individual networks.

Content processing - Making it all fit together

Up till now, we have described the compression and multiplexing part of what is describing the next generation video head-end. We have seen how the latest encoders can deliver high quality video from large high definition screens to the use of automatic re-purposing to send the same content to MobileTV screens without losing critical details. Advanced IP based statistical multiplexing has allowed encoders to be placed with the content providers, reducing operational costs whilst improving quality. This section will describe the growing importance of how content is processed before transmission and what is required from the new generation of multiplexers and stream processors.

Multiplexer turns content processor

Each network type has its own characteristics and in general fairly different in terms of how the head-end is designed to minimise the transmission cost. An IPTV head-end will stream a large number of individual Single Program Transport Stream (SPTS) whilst a satellite operator will use Multi Program Transport Stream (MPTS) with statistical multiplexing to minimise the transponder rental cost. A cable operator will typically require a large level of local program insertion. As encoders are increasingly being placed with the content providers, the multiplexer will need to evolve to process the incoming compressed streams to make them suitable for the different network types. In short, the multiplexer will need to have the capabilities of multiplexing and PSI or SI handling, conditional access insertion, video rate shaping and local program and advertising insertion.

Bit rate shaping

Transrating is the most common technique for altering the bit rate of a transport stream. This is done by performing re-quantization of each transport stream. The re-quantization process reduces the complexity within each video stream by removing information in the high frequency coefficients. This is normally done by redoing the quantization process with a harder compression level. The maximum transrating capability on compressed stream is about 25 percent but this is greatly dependent on the compression format and the incoming bitrate. A relative high incoming bitrate in MPEG-2 could in some instances achieve as much as 50 percent reduction. Transrating is normally performed as part of statistical multiplexing where a small change is performed over a large number of channels. This means that the picture degradation is minimal for each channel.

For IPTV systems that are transmitted as SPTS, the ability for statistical multiplexing is not possible so the main application for transrating comes from what is often referred to as clamping. When a video is received from the distribution link, it is often compressed for satellite transmission and will have a maximum Variable Bit Rate (VBR) rate that is higher than what is acceptable on an IP network. The transrater will here only perform bitrate reduction on the video stream when the maximum bitrate is exceeded.

Transrating has a relative low processing requirement and it is commonly possible to process large number of video stream in a single unit. This also means that the cost per channel is low compared to the other technologies.

Digital Program Insertion

Digital program insertion (DPI) is the process where a digitally compressed broadcast stream is inserted into another stream to replace the main program. This process is referred to as splicing and is used for inserting local programs or local adverts into the main national program.

This process can however be adopted to sit in the central head-end to include in this next generation architecture. If a single encoder will serve multiple networks, it is highly likely that the adverts played out will change depending on the service. An advert broadcasted on the SD network will not be the same as the one used for HD and the program line-up may not be the same. This paves the way for local insertion being performed in the central head-end to meet the growing demand for more targeted audience.

The introduction of IP for transport stream carriage in video head-end has allowed for increased design flexibility and has given way for new processing technologies to be introduced. Next generation head-end needs to be able to address an increased fragmented customer base over multiple networks with different characteristics. The video format will also vary from small mobile phones to large HD screens. By combining the best in class IP technology with advances in video processing, this is starting to become a reality. Encoders are increasingly able to deliver multi-format content. At the same time, these encoders can take advantage of technologies like RoI for dynamically adopt the picture to be able to serve the different screen types now being used. MPEG-4 has matured to the point where it is delivering on its promise to reduce bandwidth with 50 percent paving the way for the introduction of HD. Traditional multiplexers are evolving into content processors able to serve multiple networks. By combining rate shaping, program insertion and statistical multiplexing, these new generation content processors are able to handle multiple networks.

By adopting all the latest technologies now becoming available, it is possible from DTH operators to address the increasingly fragmented customer base. The future media delivery platforms will need to evolve to meet the demand for content everywhere that is expected by the new viewers.

The author of this article is Are Olafsen, Thomson, France