• Archives

What use cognitive access?

Cognitive AccessThe logic seems compelling. Spectrum is scarce and expensive but much of it is unused in any given location at any particular time. Finding a way to make use of this spectrum must therefore unlock significant value and be worth pursuing. It is this logic that appears to have sparked interest in so-called “white space devices” in the US with many major players including Microsoft, Philips, Motorola, Google and others coming together to lobby regulators and develop prototypes. Such major backing surely indicates a major market and a rosy future?


But there may be a flaw in the logic. Finding a way to make use of the spectrum only generates value if the means to do so – the cognitive access – does not result in device costs or service restrictions that are greater than the value of the spectrum liberated. Quantifying any of this is almost impossible – not least because the value of spectrum is not clear – but there are significant device costs and service restrictions when cognitive access is used.


The device costs come from the need for cognitive devices to sense whether spectrum is unused. Explaining this in detail would be an article in its own right, suffice to say here that very sensitive detectors are needed which are currently proving difficult to implement in laboratory equipment, let alone consumer devices. Of course, technology improvements or novel approaches may change all this but in many cases the issues are down to fundamental limits rather than a lack of, say, processing power.


Even more problematic may be the service restrictions. There is no guaranteed access to white space. The owner of the spectrum may start using it more intensively (or even simply transmitting unwanted data just to stop other, potential competitors, using their spectrum) or many other white space users may also try to access the spectrum. So it will be difficult to offer services that require immediate and well-understood access such as voice traffic. Indeed, any services that require infrastructure might be difficult – the infrastructure would have to scan multiple channels and would be more expensive to deploy than systems such as cellular networks. Building such a network to access what is essentially unlicensed spectrum would be a risky business. Device-to-device communications might be feasible but this can already be accomplished in unlicensed spectrum at 2.4GHz, 5GHz and other smaller bands without all the added complexity of needed to scan carefully for unused channels.


The proponents of cognitive access do not appear to have clear plans for applications. They talk more about “wireless clouds” and “build it and they will come” type ideas. Grand schemes such as the interconnectivity of a wide range of consumer electronics items are mentioned. Maybe they will turn out to be right, but they have a lot of obstacles to overcome on route to such a vision.