Just a few weeks ago Siano delivered on promises made earlier this year to deliver a chipset for more or less any form of mobile TV which comes from a broadcast network. That means its CMOS chips can receive and decode DVB-H, DAB, DAB Enhanced Packet Mode (EPM) and T-DMB. The only format missing here is the ISDB-T used in Japan.
The chips can also work with the full resolution DVB-T digital terrestrial broadcasting system that is favored in Europe and gaining traction in the Far East.
The company claims to be the only company currently delivering all of these protocols on a single chip, and its architecture is described as a set of distinct hardwired engines, surrounding a CPU core supplied by UK based ARC International.
But the market lead to date has gone to DAB chip maker Dibcom, which announced samples of a DVB-H chips at the 3GSM show in February offering T-DMB and DVB-H combined on one chip.
Since then there have been quite a few fabless silicon designers emerge all over the world chasing DVB-H and the major silicon houses such as Texas Instruments and Philips have begun to enter the race, although none actually have DVB-H silicon yet. Broadcom (see separate story) has acquired a DVB-H chipmaker this week.
We spoke to CEO Alon Ironi this week, now that his company is producing chips. ‘Our architecture incorporates a processor from ARC for housekeeping jobs, management and to accommodate changes that may happen as the standards evolve. That way we can make microcode changes and no new mask is required and no new chip run.’
‘But the critical algorithms for real time processing and number crunching are on the hardwired engines. We have the RF tuner and the demodulator in one chip, and this can identify which type of signal is receiving DVB-H or T-DMB or whatever.’
Does that mean that handset OEM customers of Siano would have to pay for Intellectual Property licenses for all of the standards that the chip supports? ‘Handset makers only pay a license for each handset that works with a particular standard. If the phone works with T-DMB then it has our microcode and an antenna for T-DMB, and its RF tuner works in spectrum set aside for that service, then it pay royalties on that standard only.’
‘Our chip is really just about tuning and demodulating the signal, it then offloads the Electronic Program Guide processing and the video codec decoding onto the host cellphone CPU, because we think that’s the right place to do the decoding,’ said Ironi.
Naturally, any new radio and processor inside a phone will use up power, and power on a cellphone is expensive real estate. ‘In power we are the number 1, far better than all of our competitors,’ he claims and has in the past placed his competition at around 40 milliWatts (mW) to 45 mW when like is compared with like.
But again this is a tight call. Both Siano and Dibcom mention 25 mW when operating in DVB-H mode, and it looks like all the serious players are heading for very similar specifications in time for bulk delivery.
In a way, there is no big reason for being first in this market. Right now the recognition these chip makers can achieve is with companies such as Crown Castle, which is a tower owner building out a network of transmitters for DVB-H in the US, and handset makers like Nokia, and this is just with prototypes.
So the fact that Texas Instruments will not deliver its DVB-H chip which is dual mode with the Japanese standard ISDB-T on the same chip until mid-2006, won’t necessarily mean that it is at a disadvantage, especially as it comes as part of its OMAP cellular chip roadmap. Philips instead rolled out a board level product for prototypes and is planning to deliver final single chip silicon in the fourth quarter.
And the other DAB chip leader, Frontier Silicon has been showing samples as recently as a month ago, neck and neck with Siano, combining both T-DMB and DVB-H on a variant of its established Kino range of DAB chips. An earlier version out in June offered just T-DMB.
Right now, with the exception of Korea, which has some S-DMB (Satellite Digital Media Broadcasting) which has very poor indoor coverage and a network that is reported to be full of holes, there are no active mobile TV services. There are an estimated 40 trials in countries as diverse as Japan, the US, Korea, Germany, Australia, Finland, UK, France, Italy and The Netherlands, but still there are many major countries that have not even yet considered if and how to go about building a broadcast network for cellular TV.
Spectrum though is proving the most serious issue for DVB-H, especially in Europe. In the US Crown Castle has landed some prime 1.67 to 1.675 GHz spectrum, and is aiming to deliver 16 channels of TV over this.
But in Europe things are not quite that simple. ‘In Europe there is a lack of UHF spectrum, so 1.45 to 1.49 GHz spectrum, which is used for DAB (Digital Audio Broadcasting) has been proposed to the DVB-H committee and they are considering it. I believe that some of the European countries, especially in Scandinavia, and perhaps in Germany, are planning to use this for DVB-H.
This is generally referred to as L-Band and is the spectrum that DAB is deployed in. Where licenses have been issued that specify which technology or service must be deployed, this is not an option for DVB-H. But elsewhere, and as Europe becomes increasingly liberal in its telecommunications regulation, it could be ‘stolen’ away from DAB.
‘And remember in May 2006 there will be a re-allocation of many of the European analog TV spectrum, to make space for new digital TV and mobile TV spectrum,’ says Ironi
In the UK, there are trials actually using DAB Enhanced Packet Mode (which supports more of a pure IP style delivery) being run by British Telecom and Virgin Mobile, and Ironi thinks this may be because of restrictions placed on DAB spectrum.
Faultline has asked in the past just how good DVB-H is compared to the other technologies, in particular the proprietary effort from Qualcomm, MediaFLO.
‘I have seen demos of all of them, except MediaFLO and we don’t support that because it is not an open international standard. They (DVB-H and T-DMB) can each achieve Quarter VGA which is 320 by 240 at 30 frames per second using up a total bit rate of between 250kbps and 384 kbps. There is a difference when you use 384kbps, when you have to read anything on the screen or watch a moving ball and if I had to advise any operator I would tell them to use the full 384kbps,’ says Ironi.
‘There has been some prototypes that didn’t have the right silicon that have used QCIF, which is 176 by 144 pixels, but I know of no DVB-H service that is going to use that resolution. They will all use QVGA.’
‘S-DMB lags these other technologies, it has a frame rate of 24 frames per second and has a weak signal indoors and needs lots of in-fill transmitters.’
Ironi also shrugs off the idea that DAB EPM is any worse an option than the other two main contenders. ‘It may take more spectrum to deliver as many channels, but companies using this can usually take more spectrum. The key here is that they can rely on existing infrastructure that was designed for DAB and use it for video.’
Faultline has long argued that one of the weaknesses of all of these video options for mobile phones is that once a program is broadcast and perhaps stored as QVGA, the entertainment can never then be played out on a full TV or PC screen, without looking vastly inferior. Effectively there is a factor of 5.4 times more pixels in a full VGA of 720 by 576 compared to a Quarter VGA of 320 by 240.
So as cell phones get more disk memory or huge amount of flash memory, there will be a tendency to download video on a PC and to use the cellphone not as a way of playing it out, but as a way of carrying it around. Various scenarios such as record at home, play on the train, record at home, view on a hotel TV, are likely to emerge, and for this at least full VGA is required.
‘I agree entirely,’ says Ironi, ‘but this is all a matter of timing. We think that for the first two years, 2006 until mid-way through 2008, that these TV phones will not have huge hard disks. There will be some memory of course, for DVR type use, but not really big disks until the second half of 2008.
‘But we see another scenario before then, where in Europe, Taiwan and Singapore (not US) both DVB-H and DVB-T will be transmitted in parallel. The DVB-T will be picked up by regulation TVs, PCs and PDA-like pocket TV players, while the phones will pick up the DVB-H.
‘But initially this will be a very low percentage of viewers. And when hard drives come out on cellphones, this will be a very small percentage again of mobile viewers. Outside of the UK, which has huge DVB-T penetration, this approach will stimulate the use of DVB-T and we see more parallel services being offered over DVB-T and DVB-H.’ Which presumably is why the Siano chip supports both.
Effectively Ironi sees a merging of mobile storage and pocket TVs post 2008, but not before, and by then usage patterns for mobile TV will be well established and slower to erode.
What of existing cellular TV which is either multicast or offered Video on Demand? ‘Services like MobiTV or any other multicast are not comparable in quality to DVB-H. Of course it depends upon the service provider. If they dedicate a good chunk of their cellular network to video, then some quality video channels can be delivered.
‘If they are happy to occupy a large number of voice channels to offer a multicast of some special event, then operators can do this. But they will not be able to support any more than a few channels before they run out of bandwidth. They might show one football match, and then later in the day offer something else at QVGA quality, but it’s not an efficient way to use the cellular signal.
‘Our own opinion is that broadcast TV services will not eliminate multicast or VoD video services. When the end user wants to zap though the channels to see what’s on they will use DVB-H or other broadcast service, when they want to see yesterday’s football goals, they will use VoD or Multicast. I do not think any operators will choose to match the bit rate of DVB-H and instead will stay at 10 to 15 frames per second and low bitrates for cellular TV.’
UK broadcast company Arquiva told Faultline a few weeks ago that an area the size of the UK, in L Band, might take 5,000 transmitters to fill, with half of them filler transmitters, to cover distribution holes including the retransmission of signals that the DVB-H signal interferes with. How is anyone going to commit to that kind of expenditure?
‘A company like Crown Castle carries out a trial and it is only when they have the commitment of a major tier one operator that they can afford to build out their network,’ said Ironi.
But can they make money from just one operator customer? ‘If they can charge between $10 and $15 a month then the network owner can create enough revenue to make money, but it depends on how this is split between the cellphone operator and broadcasting network operator and content companies.’