OK, let’s just come out and say it, DVB-H is dead. Long live DVB-T. While it may not be the first time you’ve heard that as a sentiment, we’re pretty sure that this time it is clearly, demonstrably, and totally shot as a technology.
So where’s the evidence you may ask?
Well the last drop of evidence is in the discussions we have had this week with Digital Fountain, the FEC supplier which is attacking the DVB-T market with software which can retrofit into existing DVB-T installations, to add the capability to pick up DVB-T while on the move. It has the side effect of also granting about 1dB to 2dB of improvement in in-building penetration as well. It’s not that Digital Fountain believes VB-H is dead, it’s just us at Faultline.
Marshall Porter, Director of Mobile at Digital Fountain told us that there are now moves to spread the use of DVB-T outside of Germany and Austria in handsets, specifically with LG Electronics leading the way. It is only a matter of time before an avalanche of handsets begin to adopt the terrestrial TV protocol in the 39 or so countries which have adopted DVB-T. Already phones have appeared in the Czech Republic, Spain and France, Porter said, and there are a swathe of other countries working with Digital Fountain (he won’t tell us who or where) which are investigating the company’s forward error correction Raptor codes to take DVB-T to the phone.
Most people that have seen DVB-T on a handset realize it has three major drawbacks when compared to DVB-H, the battery burns rapidly, the signal can only be viewed inside a building when near a window and the signal evaporates in a moving car or train.
The last issue is due to Doppler effects and in DVB-H this is got around by using its MPE FEC, a form of multiprotocol encapsulation. In effect an IP datagram is spread over multiple carriers, and then a table of Reed-Solomon error correcting codes is computed and sent as separate 64 column data table over a longer period of time, around 150 milliseconds. The net result is that Doppler interference can almost be cancelled out. This system anticipates roughly a 25% overhead to create this FEC.
This doesn’t exist in DVB-T but the new Raptor codes built into Digital Fountain’s Toughstream server can compensate for this with an overhead closer to 3% to 4%, and do just as good a job. The result is an improvement of around 4dB in the effectiveness of a the DVB-T signal, which will cope with Doppler effects. But is this enough?
Remember that DVB-T signals are designed to reach an external, non-moving antenna, about 10 meters from the ground. Handset antennas are mobile, move around and are around 1 meter from the ground, and in most DVB-T networks across Europe, a handset based chip with an internal antenna would fail to pick up much of a signal at all.
But a slight increase in the sending power of the DVB-T transmitters, something that is anticipated once neighboring spectrum is freed up from analog TV transmissions; along with this improved, but low overhead FEC, plus increased sensitivity in antenna design, and possibly some diversity (picks up two signals and merges them) can all help change DVB-T so that it can be picked up on a handset. Battery life has already been improved dramatically by making the chips smaller and simpler.
‘The nervousness among broadcasters is over whether or not this is backwards compatible with existing receivers that are already out there,’ said Porter, which his company is helping to overcome by getting involved with trials, some of which will become public in the run up to Christmas, and early in the new year.
The Digital Fountain position is that it makes sense to create a chip hybrid, which can receive DVB-T and DVB-H, which Siano and DiBcom, the leading chip specialists in this sector, now both have, and to get people used to watching TV with DVB-T, and put high value paid services on a new DVB-H network. But we can’t seen this happening.
‘In the past everyone has thought that in order to build a brand new network for mobile TV, it has to reach 95% of cellular users, but with something that is not charged for, the cellular companies don’t have to achieve that with DVB-T, just put it into some handsets, and let the usage grow. Once it goes beyond a certain level of handset penetration, then it should be easy to make a business case for building a DVB-H network.’ Porter said.
It’s fairly clear to us that Porter is wrong here. It will take time for DVB-T to become well penetrated on handsets, even with this new technology, let’s say at least 18 months. Given that there are the beginnings of dire warnings about the mother of all recessions about to come across the Atlantic to Europe, cellular operators will refuse to fund ANY new networks for some time to come. What they will do instead is wait until mobile broadband has filled out, and when their current networks run out of capacity, in the 18 months to 2 year time frame, look to upgrade to the next levels of HSDPA and eventually to LTE. Then internet delivered video will become the order of the day and their own Unicast 3GPP streaming businesses will be ready to grow and improve.
Until now we have been assuming that France, Germany and even Spain and Poland and the Ukraine, as well as smaller countries like Sweden and the Czech Republic, would all build DVB-H networks. After all they all have trials which were largely successful, and it is being pushed as a European standard. But if these are the very places where DVB-T usage is being tried, then that will lead to 18 month long delays, and at the end of that period the window of opportunity for DVB-H, ahead of very high speed mobile broadband, will be too short to make an investment case.
There is the possibility that broadcasters, feeling the chill of advertising weakness on their own TV stations, will feel the need to create a new route to the handset but building out DVB-H. But of course in most countries the transmission companies are now separate from the broadcasters, and the investment would have to be made by transmission companies like France’s TDF, the UK’s Arqiva and Spain’s Abertis. They won’t agree to build a new network without a) The spectrum and b) guaranteed usage payments from broadcasters and it would be a very confident broadcaster that made a long term financial commitment without the local cellcos onside.
The problem comes in the price for building such a network. At around $1 million for the equipment for one medium sized city, and another $1 million or so in truck roll, and build out, plus some bill of materials needed for each device ‘ the bill for Italy is widely cited at $200 million for a decent percentage of population coverage (85%). Similar calculations for France and Germany would go North of that figure.
The continued failure of Europeans to launch DVB-H will lead to greater and greater disillusion with the technology, and better and better alternatives. Cellular unicasting can be dramatically improved both by better codecs (we know Broadcast International is involved in some trials to bring streaming down to 50 kbps) and the emergence of newer one way data services such as next generation 14 Mbps and 28 Mbps HSPA data services. The longer it takes for someone to press the ‘invest’ button in DVB-H, the more likely that Cellcos can make do with their own core technologies.
One of the initial failures of DVB-H was that the costs of using it are mitigated only if you already have a DVB-T transmission set up. So it was always assumed that existing broadcast networks would have the lead hand. This has always made Cellcos uncomfortable.
Will a company like Qualcomm may have the nerve to go ahead and provide a service, by buying spectrum, as it has in the US, and which it holds in the UK, and absorb the costs of build out from Cellcos? Well it might, but the L Band spectrum it has bought in the UK will cost more than double to build out compared to 700 MHz spectrum. And even if it does, it just puts another nail in the coffin for DVB-H, because although MediaFLO assumes the same business model, it is a bitter rival for DVB-H.
The other major piece of evidence that came out this week was the fact that Mobile 3.0 in Germany has agreed to return the DVB-H license that it acquired to the Regulator. That puts DVB-H on ice there and the regulator wants a big powwow with all the interested parties and then to put out the license for tender yet again.
Given the bitterness in Germany and the way the Cellcos there have deliberately sabotaged the Mobile 3.0 plans, and the way Mobile 3.0 oened no dialogue with the Cellcos, it seems likely that any future auction of that spectrum, if it is allocated solely for DVB-H, will attract few, if any, bidders.
Finally, speaking to Porter, he made it clear that even if the rest of Europe goes down the route of strengthening the DVB-T signal with FEC and increased power, the UK will not benefit. ‘A channel on one of the UK multiplexes now costs £10 million, so the extra 3% or 4% that we need to make it robust enough to reach a handset will cost too much.’
That leaves the UK waiting for pay TV operator Sky to either bankroll the Qualcomm effort; or for one of the local Cellcos like Vodafone to beef up its unicast service with new codecs; or for the IP Wireless experiment in London between T-Mobile and Orange, to form the basis of mobile TV there, with no DVB-H or DVB-T in sight. Another possibility is that Alcatel Lucent and its satellite partners will make a bid for the UK with its DVB-SH system, said to have a 40% deployment cost advantage over DVB-H.