Streaming audio services can provide reasonable quality at 30kb/s to 40kb/s. For example, BBC Radio offers streaming of its live programs, illustrating the fact that radio-on-demand has become a reality.
TV broadcasters do not currently consider the Internet to be a suitable delivery mechanism for their services because most users access the Internet at speeds of 56kb/s or less. Delivering good quality video and audio at such low bit rates is currently not practical. In fact, it is debatable whether ‘video’ is an appropriate term for blurred pictures that are not much larger than a postage stamp, and are refreshed only three or four times per second.
On the other hand, streaming audio services can provide reasonable quality at 30kb/s to 40kb/s. BBC Radio offers streaming of its live programs, plus the opportunity to listen to an impressive range of radio programs broadcast over the previous seven days. This example shows that radio-on-demand has become a reality, but when will we have equivalent services offering TV-on-demand via the Internet? Broadband delivery to consumers promises to solve the bandwidth problem, but will it permit broadcasters to use the Internet as a new delivery mechanism for their TV broadcasts, and will it allow video-on-demand services?
What is broadband?
Broadband services are typically delivered using cable modem or ADSL technologies. The term ‘broadband’ is now widely used to describe services running at 512kb/s or even 256kb/s. It is interesting to reflect that in the early 1990s, 2Mb/s was considered to be narrowband. At that time, many telecom companies expected that their customers would soon be demanding fiber-to-the-home (FTTH) services offering much higher bit rates. Consequently, they were reluctant to invest in so-called ‘intermediate technologies’ such as ADSL.
In the mid-1990s, video-on-demand (VoD) was expected to be the killer application that would justify broadband delivery to homes. As the MPEG-1 compression scheme could deliver VHS-quality video at about 1.5Mb/s, network operators realized that delivery of 2Mb/s via cable networks or ADSL would permit VoD services. This concept of VoD was eventually abandoned when it became clear that consumers would have to rent an unrealistically high number of movies per month to justify the cost of providing the broadband networks to individual homes. The competitors in the video rental stores had no such constraints; in essence, their customers were responsible for delivery of the movie, either by driving or walking to and from the store.
By the late 1990s, the Internet had become the new killer application. Nobody was quite sure what the Internet would do, but broadband could obviously offer high-speed access to the Internet, and those of us who now have such broadband access would not willingly return to dial-up connections. The fact that broadband is ‘always-on’ is almost as important as the improved speed.
Delivery of video at 300kb/s can produce acceptable or even good quality when the picture is relatively small, such as 240 x 180 pixels, and when the pictures are not very demanding, such as talking heads or material shot on film. However, 300kb/s is far from adequate for coverage of fast-moving sports. This is not surprising since the MPEG-2 compression scheme demands 4Mb/s to 5Mb/s for standard-definition digital TV services. New techniques in video compression will certainly reduce the required bit rate, but 300kb/s will remain a very difficult target for high-quality full-screen video.
A related problem is that computer users tend to be demanding in terms of picture quality because they are close to the screen; the viewing distance is typically less than two times picture height (2H) for computer displays, instead of 6H for TV sets. Whereas VHS quality was considered to be the reference in the mid-1990s, the public now expects much better quality. The reference quality is now that of DVDs, which are coded at 8Mb/s to10Mb/s, with the parameters being adjusted on a scene-by-scene basis to give optimum quality. This trend towards higher quality seems set to continue with the arrival of HDTV broadcasts and the promise of HD material on the next generation of DVDs.
The solution might be a mixture of better compression schemes and higher bit rates, such as 2Mb/s, on broadband delivery systems. Nevertheless, delivering HDTV at 2Mb/s will be a formidable challenge.
It is tempting to imagine that delivery of audio and video services via the Internet is just like broadcasting, but only better because the Internet supports on-demand services and is available throughout the world. Unfortunately, the reality is rather different. Many broadcasters are amazed to discover that there is a finite limit to the number of simultaneous users who can receive their audio or video services via the Internet. Some broadcasters’ Internet servers can accommodate 10,000 simultaneous users, but many have a limit of 500 or less. The problem is that each individual user needs to receive a continuous dedicated stream of data, unlike broadcasting where the transmitted signal can be accessed by an unlimited number of users.
If your audio and video services delivered over the Internet become very successful, you will have to pay for more hardware, such as servers, and for greater bandwidth. Consider an example where a broadcaster offers on-demand access to material previously broadcast live over the past week.
Some effort is needed to re-author the content so that it is suitable for delivery over the Internet, but this would be relatively cheap. Let us assume that the incremental cost of production would be E50,000 per year and that the cost of delivery via the Internet would be another E50,000 per year. As the delivery cost is proportional to the size of the audience, increasing the audience by a factor of 100 would increase the total annual cost from E100,000 to E10 million. This example illustrates how you can be the victim of your own success on the Internet.
It is instructive to compare this example with that of broadcasting, where the costs of broadcast distribution are a small part, typically 2 percent to 5 percent of the overall budget and, more importantly, independent of the number of users. Broadcasting is a ‘public good’ because the marginal cost of extra listeners or viewers is zero, and it has been so successful because the costs of production and delivery are shared among millions of people.
Delivery via the Internet could be successful economically if the users were prepared to pay some form of subscription fee, perhaps per minute or per month, for access to the services. How many people will be prepared to pay for services delivered via the Internet if they already have direct access to similar or identical free-to-air broadcasts? It is doubtful whether subscription services would be viable, except for niche markets such as expatriates, who are often keen to watch or listen to broadcasts from their national broadcasters. However, the global reach of the Internet brings yet another problem: Many broadcasters do not have the right to transmit their services around the world.
At first sight, broadband seems attractive for TV broadcasters because it can enable delivery of on-demand services to consumers. Closer examination reveals two major problems. First, broadband is not broad enough. Secondly, delivery of broadband services could become very expensive for broadcasters. Although improved video compression might be a partial solution to the first problem, there is no obvious solution to economic problem.
Philip Laven is technical director of the European Broadcasting Union.