Setting radio free

The Open Future for Wireless Communications conference in Cambridge, organised by the Communications Innovation Institute in conjunction with the Cambridge-MIT Institute, saw technologists, regulators, researchers, network operators, equipment and chip manufacturers get together to discuss the future of radio regulation.

The event, held late last month, included a discussion on ways to add new services to the radio spectrum. One was the freeing up of existing bands — spectrum re-farming — by making existing services use more efficient technology and reallocating those frequencies which are either lightly or no longer used. Another involves the creation of underlay networks — wideband services that run alongside existing occupants without mutual interference. Finally, overlay technology takes existing networks and combines them behind a common interface — improving reliability and quality of service for the users to the point where new services can be introduced.

All of the above options require changes in the way radio is regulated. To date, spectrum re-farming has been the most popular option as it can be done piecemeal and reduces the number of people affected at any one time. For example, there has been a move from analogue to more efficient digital services with mobile phones, public service users such as police, and broadcast radio and television.

And there's no doubt that more can be done here. Professor William Webb, head of research and development at Ofcom, said that "Nobody thinks that they have unused spectrum, but we do". A spectrum survey at three sites, Heathrow, Central London and Cambridge, showed that half the spectrum between 100MHz and 1 GHz was unoccupied, he reported — "a pessimistic figure, but not too far off". The unused portion was mostly military. "There's a deep suspicion that there's more than enough spectrum for everyone."

Underlay networks are the most technically precocious option, and involve massively spread spectrum technologies like Ultrawideband (UWB) or opportunistic adaptive systems like cognitive radios that sense the spectrum around them and seek out unused portions. Here, the question is how much underlay networks will interfere with existing services. Simon Pike, chief engineer, Regulatory and Spectrum at Vodafone, was cautiously pessimistic. "UWB interference depends greatly on the type and location of the systems affected," he says," but it's unlikely to be avoided completely. "UWB could prevent the main licensed user of a band from improving the spectral efficiency of his system, and could restrict future developments that could have even greater benefits", he said. He was also unimpressed with cognitive radios which can sense their environment and location and then alter power, frequency, modulation in order to dynamically reuse available spectrum. "Cognitive radios can only identify transmitters, not receivers, so you get the hidden terminal problem [where one side of a link is inaudible to a sensor because it's too far away or shielded]. They're also potentially counter to liberalisation, as cognitive radios can only respond to signals they know about. Once a cognitive system is deployed, it's hard to change the use of a band to something those radios don't recognise".

Overlay networks have the advantage that they use existing systems. While regulatory changes will be required to make best use of them — licensees are very constrained in what they can do with their systems, regardless of changes this will make in the actual spectrum used — these may have minimal impact on other users. Cambridge researchers have been working on PROTON — a policy-based system for roaming transparently over overlaid networks. This tackles the problem of how a connection can seamlessly switch between networks with different characteristics — cellular, LAN, wireless WAN — ideally, aggregating as many as are available to maximise throughput and quality of service. The problem boils down to three distinct areas — deciding what to do, doing it and changing behaviour as the environment changes.

It's not good enough to have hard-wired rules such as mobile phones use to change between cells, switching to the strongest signal or picking the highest bandwidth; instead, the nodes on the network must analyse their surroundings over time and decide how to connect based on a policy of what works best under various conditions. So, a connection on GPRS that detects it's moving into a Wi-Fi zone might decide to negotiate a connection ahead of time, because it knows that switching over at the last minute involves a break of several seconds. Cambridge has a live wireless IPv6-based testbed for PROTON, using a combination of GPRS and 802.11b and whatever else they can plug in. Whatever technologies go together to make the fourth generation of mobile services — 4G, if you will — the thinking runs, there'll be many different networks in many different combinations; the big challenge will be to make them work together without any input from the user.

Regulators face an impossible task. For any territory, there are a range of incumbents already conducting business in spectrum they consider theirs — in some cases, having paid very large amounts of money for the privilege. There are also newcomers itching to get going, who consider it unfair that they are denied access to the airwaves because they have technologies that don't fit into the old idea that services had to be restricted to one band and one way of doing things. The incumbents have legitimate technical concerns that the new services might interfere with their existing expensive infrastructure and mess up their business models. They also have commercial concerns that the newcomers might be more efficient and cheaper, in ways that the incumbents would be prevented from emulating due to the terms of their licence. Also, they're happy to see competitors strangled at birth — as all companies are. That translates to an understandable willingness to paint the bleakest picture of the consequences of allowing new and less-regulated services.