Qualcomm has just paid Ofcom £8.3m for a 40MHz-wide chunk of the "L-band".
The licence covers the whole country, has no termination date and has no restrictions or obligations as to technology or rollout. The company can do pretty well what it likes, where it likes, when it likes. But Qualcomm hasn't said much about its motives. "Acquiring this spectrum will enable us to develop, test and explore a variety of innovative wireless services and technologies that will benefit European consumers and the wireless industry as a whole," states the press release. It's up to us to try and work out Qualcomm's plan.
The L-band is a rather peculiar beast. Until the 1970s, technology capable of using it was too expensive for consumers, so it bears the legacy of plenty of interesting professional and esoteric uses. Roughly bracketed at either end by the mobile phone bands at 900MHz and 1800MHz, it contains a mixed bag of military and civil services, including radio amateurs bouncing signals off the moon from Stockton-on-Tees. GPS and Galileo live there, as does digital audio broadcasting, both terrestrial and satellite, as well as space-based ground-measuring radar, astronomy and just about anything else you can think of.
These days, it's the natural beneficiary of all the work that's been done in making cheap and effective mobile-phone technology. With a wavelength of around 20cm, it's easy to make portable devices that transmit and receive efficiently, and the electronics required are available off the shelf for a matter of pennies. It gets through walls better than Wi-Fi, it isn't attenuated that badly by the atmosphere, so you can get decent coverage from well-sited base stations, and, with 40MHz to play with, Qualcomm can easily deliver a couple of hundred megabits per second over whatever it is that it chooses to do.
In this case, it's safe to take Qualcomm at its word: it's got some new ideas and it wants to find out whether they work. With the spectrum auction win, it's effectively bought one of the world's biggest open-air radio-frequency laboratories — at an absolute knock-down price. The UK contains just about every sort of radio environment, from dense masses of urban canyons — the great phrase that radio engineers use to describe skyscraper-studded financial districts — to the bleakest of rural vastnesses, via the prairies of suburbia. Plus, just as importantly, these physical environments are matched by just as rich a mix of economic environments. The L-band has the attributes of just about any other band that might be used to deliver terrestrial wireless services, which means that any work Qualcomm does will be useful, no matter how or where the eventual commercialisation happens. It's the perfect petri dish in which to grow novel concepts.
This seems like a much better use of the band than the alternatives. One of these, detailed by Ovum, is that Qualcomm would roll out an L-band mobile-TV service based on its MediaFLO technology, with the intention of developing the handset, operator and content sides in preparation for the availability of better spectrum come the analogue switch-off in 2012. The trouble with this idea is the huge investment by partners in something with no very good chance of turning a buck in its own right — and, in the West at least, mobile TV is looking like a dangerous gamble.
Given the huge and increasing pressures on broadcast TV from online services, putting a big bet on replicating it in handheld form with a launch date four or five years hence seems foolish at best (not that foolish bets are unknown in wireless). Plus, MediaFLO was concocted five years ago, before other people had tried and mostly failed to roll out mobile TV; the latest and biggest hopeful is AT&T, which last week launched its MediaFLO service in the US. The success or failure of that will have a much bigger impact on the popularity of the technology than anything on the UK L-band. By now, MediaFLO — the FLO stands for "forward link only"; it's a one-way street — seems a bad match for a Web 2.0 world.
It's much more interesting to think about what could be done differently...
...in the environment we can expect in 2012. That means thinking big. But how big?
The next decade will see 3G/HSDPA offering a lot of services very cheaply, and the joy of IP is that you can create new services over existing infrastructure very easily. This will be even more true in that time span; by then, handhelds of all types will be easier to develop for and deploy to. Whatever Qualcomm comes up with, it will have to be strong where the 3G world is weak. It can't afford to be another WiMax, unable to find a strong differentiator to set it apart from the competition.
The weakest link in 3G is operators — big, fat concerns, hugely dependent on ripping the punters off by charging stratospheric overheads for simple services. A text message costs an operator nothing to carry, yet costs users four times more per byte than Nasa spends talking to Hubble. The operators' biggest worry is that it is hard to see how to carry margins like that across to the way consumers want to use mobile data — grabbing and interacting with internet content — and that, more than any technical concern, limits what 3G will be doing in 2012.
All of this makes the most interesting problem one of bypassing the operators altogether. That means creating a big, open market where infrastructure is cheap, bandwidth plentiful, and producing and using devices is as easy as it is with Wi-Fi gadgets (WiMax was going to fill this role, once upon a time). And there, the biggest problems are base stations and backhaul. How do you roll out a nationwide service without thousands of base stations and gigabits of backhaul distributed expensively to all of them?
Wide-area mesh doesn't work well with Wi-Fi: the frequency is too high, the power too low, the band too congested. You need somewhere where you can use more power, where there's no interference, where you get more reach and more freedom in setting the protocols. Something like an unfettered 40MHz on L-band would be almost perfect.
It is a huge leap of imagination to see a successful end-user mesh network, where one consumer has to be within radio range of another to get connected. Getting to critical mass will need the creation of something that looks remarkably like an operator, to provide a skeleton on which the flesh of the mesh can grow, even if the business model is very different — think of it as being more like a discounted handset, where the initial operator infrastructure is provided free, in expectation of revenues to come.
But there is much to recommend it: unlike existing configurations, adding more nodes to a mesh actually increases throughput for everyone and, if your business model involves selling as many chips as possible, then this is a very interesting model indeed. Unlike existing 3G systems, your economics get better the more users attach and consume data. They're doing the work for you. It's not a model you can explore in an ordinary lab: you have to get stuff out there to have a chance of working through the many issues.
All these factors don't add up to a compelling case for Qualcomm using the UK as a mesh testbed. The company has a huge investment in making MediaFLO work, and it's dangerous to underestimate the internal pressures such factors bring to bear in quashing diverse thinking within an organisation. It's not clear that the company has the right intellectual property to capitalise on such a system and, if it doesn't, whether it could develop it. Yet it makes more sense than creating yet another MediaFLO demonstration system or yet another "me-too" wireless data service. It would be a very big win indeed.
Whatever happens, I envy Qualcomm. In its UK L-band segment, it has a blank canvas on which to paint a completely new picture. The industry's expecting Vettriano; let's hold out for Picasso.