The technological barriers of delivering quality IPTV signals have been overcome, leaving the way clear for operators to roll-out services, writes Thierry Boudard.
The idea of IPTV - delivering a "broadcast" television service over a broadband infrastructure - has been around for a while now, but it is just starting to enter the "take off" phase. Globally just 2% of households receive IPTV, according to the latest research from Screen Digest and the IABM. How can technological advances boost take-up?
The biggest challenge, of course, is to deliver a quality of service that audiences expect. In many of the biggest markets, consumers can choose a broad range of television services from terrestrial broadcasters, satellite or cable. If these deliver the choice and quality they demand - including a good selection of high definition channels - there is no incentive to move to IPTV.
The internet protocol was developed to carry packets of data, and it is an amazingly successful way of carrying emails, web pages and documents. The difference between these and a live television signal is that IPTV demands real-time performance. To the typical consumer, there is no real difference between a web page taking half a second to load or two seconds, but television demands a new frame every 40 milliseconds, come what may.
The IPTV system divides the compressed television bitstream into packets of data and adds headers and other system level information. These are multiplexed into other data services and streamed out over the IP network to the receiver. Along the way, there are plenty of reasons why some packets get delayed relative to others: even if they leave the headend on a regular clock, they will arrive at the set-top box with varying delays.
To cope with this, the IPTV set-top box has to have an input buffer that can accept the packets as they arrive but push them into the decoder back at the regular clock interval. This is fine so long as the delays stay within reasonable limits.
If some packets are subject to a long delay, then the buffer will empty, the decoder will have no picture or sound information, and the screen will freeze. On the other hand, if there is a rush of packets with very little delay then the buffer will overflow, resulting in some packets being lost. Again, this means visible or audible picture degradation.
One solution is to make a bigger buffer, but high-speed memory is expensive and consumes power, so this adds to the cost of the set-top box and its energy consumption. It also adds to the latency: when everything is working normally, the packets still have to be clocked through the buffer - add memory and it takes longer to clock it through. This may not be a problem when watching a programme, but it adds to the channel switching time.
These problems can be minimised, but they do depend on a close relationship between the headend and the set-top box.
First is a technique for minimising clock jitter through the system, thereby reducing the causes of the problem. Second is an advanced forward error correction scheme which attempts to mask jitter errors by giving the set-top box a means of predicting what was in any lost packets.
The third is a feedback mechanism that allows the set-top box to request retransmission of any lost packets. This is, of course, the most secure way of maintaining the quality of service, but it has its downside, again because television is a realtime service. There is no point in requesting the retransmission of a lost packet if its place in the bitstream will have been decoded by the time it arrives.
The request and the response has to happen before the relevant point in the buffer reaches the output. Without intelligent algorithms and tight integration between the headend and the set-top box, realtime packet retransmission can cause more problems than it solves.
The second set of challenges is in delivering high quality television channels over IPTV. Today, 88% of IPTV subscribers worldwide are currently receiving services based on MPEG-2 compression. HD over MPEG-2 requires around 15Mb/s, which is more that the typical ADSL2+ circuits to the home can offer, even without other data services sharing the same pipe.
There are two parts to the solution of this problem: reduce the bitrate demand and increase the bandwidth.
First, all IPTV services should be looking to MPEG-4 as its compression standard. This is a more recent development, taking all the experience of MPEG-2 and building on it more powerful compression algorithms. It can now be demonstrated to deliver excellent HD quality in as little as 4Mb/s.
It is clear, though, that to achieve this level of compression efficiency requires very much more processing power than MPEG-2.
Telcos keen to achieve good market penetration with IPTV will also have to consider increasing the raw bandwidth available. In some places - new build apartment blocks, for example - it may be practical to run fibre to each individual home, giving vast amounts of bandwidth.
In more established areas, the cost of installing fibre to the home may be seen as prohibitive, but fibre to the cabinet (the distribution box on the street) could be practical. From there, the copper connection to each individual home could be just a hundred metres or so, which would support speeds of 15 to 20 Mb/s - more than enough for HD, multiple IPTV channels, voice over IP and other data applications to run concurrently.
Telcos now launching FttX look at IPTV as the solution for such differentiating services. Here the issue is not to adapt IPTV services to limited bandwidths, but on the contrary to use as much of the available bandwidth as possible, both downstream and upstream. This the newest challenge for IPTV - and definitely an exciting one.
Even fibre to the Home is going to require an investment from the telco, which will then need to ensure that it sees an adequate return on its investment. The standard measure is the ARPU, the average return per user, and the payback for the telco will need IPTV, bundled with other services, to drive up the ARPU.
While video on demand is always quoted as the killer application for IPTV, currently operators find that it provides less than 20% of their revenue. The solution is to make VoD more attractive to the consumer by pushing them to use it. This, too, requires some integration between the headend, the video service platform and the set-top box.
The first step is to make the set-top box's home screen a portal into the available content. Typically, set-top boxes use elementary HTML to generate the on-screen programme guide. By implementing Flash in the STB the user experience can be much more dynamic. It also allows the telco to insert advertising at this level, for which it will receive all of the revenue rather than sharing it with a broadcaster.
The second, more advanced step is to learn the consumer's preferences through the legacy information we get from the CRM (geodemographical data for example) and also through the user's video consumption to build a personalised television service for that individual consumer. If, for example, an individual household regularly watches programmes like Hannah Montana on a Saturday afternoon, the welcome screen might put one of the High School Musical films - pay per view on demand - prominently on the start-up screen.
Developing a preference knowledge base is well established in web applications - the online retailer Amazon, for example, always suggests purchases whenever you log on - and the same intelligence can be incorporated into an IPTV service, given a sufficiently powerful set-top box and an open video service platform.
IPTV - as a standalone service or as a hybrid in combination with a terrestrial or satellite operator - has the potential to deliver a very attractive package to consumers, and to drive up ARPU for the telco through on demand and pay per view content and highly targeted advertising. All the technical solutions are now in place, and it is up to the bold operator to put together a lucrative service.
Thierry Boudard is director of Global IPTV Solutions, Thomson.