Google, Microsoft and their friends in the White Spaces Coalition believe that they have won the war to bring cheap flat rate broadband to the US, a move which will further their plans to dominate internet advertising and PC platforms. But as the dust settles it seems hard to work out if they have really won any kind of victory at all, more than an early minor skirmish.
In fact we feel that America appears to have got itself into a bit of a mess over its White Spaces strategy which the FCC pushed through in time for the Presidential election, a move which upon analysis is starting to look to us increasingly futile. The FCC announced the order on November 4th allowing the use of UHF buffer zones, originally laid out to prevent TV channels interfering with one another, to be used for the delivery of new broadband wireless services, which have been described as ‘Wi-Fi on steroids’.
It is true that radios which could harness a complete 6MHz TV channel, to the best ability of existing wireless technology might deliver something close to 3 Mbps per MHz, or over 18 Mbps (the theoretical maximum is 19.4 Mbps), shared between local customers. However that is only true with certain modulations and in order to reach longer distances this will rely on higher power than the FCC has prescribed. Just how much bandwidth can be delivered and how far, in a real world situation, is an open question, depending on a lot of technical decisions.
So far the IEEE 802.22 working group on Wireless Regional Area Networks is planning to address this technological issue, and posits a cognitive radio and Physical Layer spec based on OFDMA and a set of policies and procedures for operation in the TV bands between 54 MHz and 862 MHz.
OFDMA is a multi-user version of OFDM, the multi-carrier organization used for most digital TV standards (all but the US standard) some of the W-Fi standard, WiMAX and mobile TV standards such as DVB-H rely on OFDM. The difference is that an extra layer of logic is used to assign sub-carriers to particular users. Usual OFDM rules apply such as changing those channels around periodically and layering data bits across multiple channels in order to create robustness against fading and interference. We presume the use of a robust error correction system will also come into play.
The Media Access Control put forward will rely on having frames and superframes, which control collections of frames, which leads to two classes of headers. These new headers are transmitted by base stations in all the channels which it is allowed to use and it is these which are picked up by items of customer premises equipment (CPE) so that it knows both what spectrum the spare channels are in which it will use to transmit and so that it also knows which of the sub-channels is already allocated to another user.
The standard also hopes to get around the 6 MHz bandwidth limits by using channel bonding, much the same way that DOCSIS 3.0 allows multiple non-contiguous channels to be used if they are available in cable systems.
There are problems in this scenario everywhere, in terms of technical feasibility, political likelihood, and economic limitations.
The first point to consider is that even if IEEE 802.22 comes up with a way to organize and share this spare spectrum, will the chips that are built for its use have the same universal appeal as Wi-Fi. In order for that to happen more than just the US needs to make the same kind of regulator order as the FCC has done.
While there are some countries with a patchwork quilt of spectrum, such as Italy, there are very few other countries that have the same history as the US in terms of the haphazard way in which it has allocated its spectrum.
So often the US is a pioneer and as always they end up with arrows in their back, as we know. The US allocated much of its TV spectrum in the 50s, well before other countries had got around to it. As such it decided that what it needed was to offer one spectrum in one state, and in a neighboring state make sure that same spectrum wasn’t used at all so that no interference between the two licenses was possible. This is quite wasteful, and in order to create as much spectrum as possible, no guardband was allocated.
The rest of the world, with the benefit of hindsight and the US experience, and the further development of technology, for the most part, decided to allocate national licenses, so that a single frequency TV network could be created end to end. Sure this created some problems in places like Europe where many smaller countries are squeezed together, but only at the borders, something that could usually be catered for by special border arrangements, such as using less powerful signals there.
As these countries have moved to digital TV they have tended, wisely or not, to allocate SFNs which run multiple channels across a multiplex of channels in what used to be a single 6 or 8 MHz analog TV. Most technologies get 6 full SD channels into one multiplex (see diagram below). These arrangements, such as DVB-T, can benefit from statistical multiplexing, which allows one channel in the multiplex to borrow data allocation not needed by another, and squeeze about 40% more programming into the same space.
The problem with countries that this approach is that they will never have a patchwork of spectrum allocations which are different from territory to territory, and as a result they won’t have any likely use for the White Spaces technology.
This was never true with Wi-Fi, where the unlicensed spectrum allocations which it uses are similar all over the world.
That definitely makes it uneconomic for a chip maker to decide to enter this marketplace. That’s not to say that no-one will, but that the pricing on the chips, if only sold in the US will need to be higher, and as a result, in order to make them appealing in financial terms, key elements such as channel bonding may be left out by some manufacturers, in order to keep the Bill of Materials down as low as possible. All that potential suppliers can really do is keep up with the 802.22 committee which is trying to address this, and try to drive a series of principles though the standard that might make sense in other parts of the world. Obviously in third world countries which have not yet completely allocated their UHF spectrum, such components may make sense.
Another blockage in terms of economic feasibility comes from the political dimension. If the FCC changes or improves any of the conditions that it put out in its White Spaces order, then any effort which a chip business goes through to develop chips may become wasted, and it makes no sense for a chip designer to consider putting its ideas into silicon until it has a glimpse of the new administration, which will potentially change the makeup of the FCC, which in turn may listen to different power groupings in this debate.
In other words, the ruling may be made, but the lobbying is not over.
We approached a number of likely chip companies this week to tell us what they thought would happen, but very few of them even came back to us, mostly because this is such a contentious issue. One company may want to push in one direction, but won’t want to upset another constituency that may become a customer down the road.
About the only company that was happy speaking to use was Qualcomm, though over the coming weeks we hope to bring the views of one of more other interested parties.
Dean Brenner, Qualcomm’s VP Government Affairs said, ‘In 1952 the TV table of assignments was established which meant that certain territories did not have certain TV channels in use. These are now vacant channels used as buffer zones to avoid interference. There are also areas in which no one applied for TV channels.
‘There is no guardband in TV, and the 1st adjacent channel to a TV station remains a special problem in terms of interference and there are special rules relating to adjacent channels.’
Brenner is referring to the fact that fixed services and portable devices operate under slightly different rules in the ruling. Fixed networks are not allowed to operate on the first adjacent channels, and portable devices have to operate in slightly less power when in a channel adjacent to a TV station. This is because fixed services will be able to use a database (once it is created) of which channels are being broadcast near the antenna, whereas portable devices have to rely on sensing the network.
Brenner made no bones about putting what has been a consistent Qualcomm position, ‘We think this spectrum should be licensed and auctioned,’ he said
‘We were one of the group that urged the FCC to step back and took a deeper look back in 2006 with a further consultation. The CTIA agreed.’
‘At present a fixed services is now allowed to go to 4 watts and have a transmit antenna not higher than 30 meters. If this had been auctioned, then the transmitters could have been allowed to broadcast at 12,000 watts.’
This would have led to significant coverage in rural areas, with far fewer antennas. Qualcomm’s own mobile TV service MediaFLO operates with 50 kW antennas which transmits at least 25 kilometers.
The personal portable devices under the ruling can operate on the adjacent channel at 40 mW, and if in the next channel over in 100 mW.
‘Also,’ asks Brenner, ‘Who is going to build and maintain the database of protected access?’ and he goes on to highlight issues here such as the need to include cable headends in the database, because currently these are not regulated by the FCC and there are around 8,000 of them in the US. They each have to receive the local TV channels in order to re-broadcast them over cable to satisfy their must carry requirements for local content.
‘The idea is that there will be an obligation to declare and register, to get on that database, if you don’t want someone to interfere with the signal in your area.’
‘There are a lot of rural homes that currently cannot receive local TV channel broadcasts at all, and TV translators will need to be on the database,’ pointed out Brenner.
A TV translator is a device which receives a TV channel in one spectrum and outputs it in another, for a different region. These are often on the edge of the transmitter range and therefore vulnerable. There are also boosters which do this in the same spectrum. Brenner’s point is that the task of creating a database of protected areas is not trivial and the harder it is, the more it will cost and the less anyone would want to take on the task.
It was during trials before the order that Motorola put forward the idea of a such database which would work with GPS in the fixed transmitter devices, so that the service would not be able to interfere with existing signals.
‘In Wi-Fi the issue is range. In White spaces the issue is bandwidth, 6 MHz will not give enough bandwidth for a Wi-Fi style service,’ he said. And he has a point. We are used to Wi-Fi sharing out 54 Mbps or 100 Mbps among crowded broadband users in small metropolitan areas, like a station, an airport or at a conference. But this new White Space ‘Wi-Fi on steroids’ will have the range to take in 100s, if not 1,000s of users, and will therefore appear parsimonious when sharing out bandwidth. Of course chip makers and the 802.22 committee might create the ability to bond multiple spectrum slices together, but that will also make chips more expensive
Also in many parts of the US there are only a few channels that are available in dense urban areas that are hotspots for TV delivery. In Alaska it is thought that some 75% of the UHF VHF TV bands between 54 and 862 MHz are available to use. But in denser urban populations it is far less with some of the channels, for instance from 14 to 20 being used for Public Safety applications.
‘We are doing an internal analysis now about what is available’ and upon hearing our suggestion that most markets will have 30% of this spectrum available he says this figure is simply ‘too high.’
‘We don’t know if a new administration will change the FCC ruling, but we know there will be efforts on all sides to try to change it. We would expect the FCC may want to make further clarifications and changes,’ concluded Brenner.
And then we come down to the issue of technical feasibility. It is a simple truth that none of the devices as they were tested by the FCC completed the tests to the point where manufacturers could go out and build the devices tomorrow. Motorola and Philips were both considered leaders here over Microsoft and Google, but still the technical committees of the FCC only used the trials to establish that the trick was feasible, not that it had been adequately demonstrated.
One opponent of the FCC order has been the Association for Maximum Service Television. This broadcaster friendly lobby said in a filing that the FCC proposal to detect signals only down to -114 dBm, equates to setting a smoke detector to only be able to detect a raging fire.
One of the biggest issues isn’t about TV at all, but about wireless microphones which use the same spectrum at live music events, TV and radio studios and even in schools and churches all over the country. The reason for the -114 dBm sensing capability being in the portable sensing white spaces devices, was so they could sense this equipment and avoid the frequencies they are using.
Wireless microphone manufacturer Shure has already begun offering a rebate program of up to $1,000 for the trade-in of Shure 700MHz frequency band (698 to 806MHz) wireless systems in order to buy those that work in different spectrum when the digital transition happens.
A spokesman for Shure said at the time ‘We’ve known for some time that the 700MHz band would be reallocated for new services following the DTV transition on February 18, 2009. Although a final decision from the FCC is still pending, we felt the need to assure our customers now that we will take care of them regardless of the outcome.’
Shure has also gone on record that it believes that for the rest of the VHF and UHF spectrum, spectrum-sensing alone will lead to a disaster for the use of wireless microphones, and is irked that its customers’ fate has been at the back of the FCC agenda all along. It accuses the FCC of burying the data at the back of the report and because individual tests which demonstrated obvious interference, were not included, concludes that the FCC has completely failed to make a case that future white space devices won’t interfere with wireless microphone use.
It says that the FCC decision will greatly complicate the lives of wireless microphone users across the United States and negatively affect tens of millions of Americans listening to live and broadcast events. And if that’s not fighting talk then we don’t know what is.
During the FCC tests, the white spaces devices which were used were clearly all bulky and pre-chip designs, so chip designs have got to be completed, a run of chips made, then these have to be tested and potentially adjusted, and finally another chip volume run carried out and physical equipment designs need to be commissioned and constructed.
Given the likely liquidity of the target design, caused by political uncertainty, none of this is going to be committed to prior to some kind of confirming statement by a newly formed FCC. All of which means it is unlikely that any complying device can get itself certified within two years.
If you add to this uncertainty with statements from the FCC about allowing devices to rely solely on spectrum sensing as long as they are subject to a more rigorous approval process and the fact that the approvals process has to be defined, a certification put in place, not to mention the matter of appointing someone to manage the database of protected access and all the uncertainty that the PC industry is about to go through in the current recession ‘ then you have a recipe for procrastination. The first aggressive PC move might trigger other efforts, but when will that be? Dell said it will definitely make PCs that do this, but talk is cheap.
All of which begs the question about just why the order was rushed through in the current administration. If the FCC chairman had realized all of this perhaps he would not have bothered.
Another aspect slowing down the fixed access design is that the FCC said it might allow higher powered unlicensed operations in rural areas. Well are they or aren’t they? In which case that has design considerations for fixed equipment and any equipment that wants to send data to these fixed stations.
Right now it is hard to see how all of the positive noises that the White Spaces Coalition has been making can convert into concrete action, unless they wish to throw money away in pursuit of what is likely to become a moving target. In the current times chip makers are unlikely to throw any cash at this sort of problem, unless they are underwritten by the corporate treasure troves of Microsoft and Google, neither one of which is scared of losing $millions in pursuit of a goal. But even they can’t break the rules of both politics and physics at the same time, and we would expect real White Spaces design activity to go quiet for a considerable time to come.