White Space: The Next Big Thing in networks
Among the mystery pub antennas of the Silicon Fens
Analysis The white spaces are filling up around Cambridge, so El Reg went up to the Fens to talk to the companies responsible about what white space is for ... and why they're sticking their masts on pubs.
Ofcom issued an experimental white space licence permitting low-power transmissions within an 80MHz-wide band around Cambridge in June, and since then local companies have been busy deploying white space radios for experimentation. We dropped in to see three of them, and discovered some very divergent ideas about how white space will be used, though there was a general consensus that Local TV was just a bad idea.
Two of the companies we visited are already running white space links around 6km, and getting a reliable 3Mb/sec in an 8MHz channel. The third has covered the whole city in a white space network, intended to connect machines to machines and showcase the company's proposed standard for such communications.
White space radios operate in frequencies which are reserved for TV transmissions, but aren't being used locally thanks to the hugely overpowered radios used to broadcast television. A transmitter in London might reach to Cambridge (Crystal Palace does, just) so TV transmitters can't use the same frequency, but short-range devices can, and short range can be pretty long.
As no one in Cambridge is using the Crystal Palace transmissions to watch TV, there's no chance of those devices knocking out local TV reception, as long as they know where they are. The weak signal from the Palace can safely be ignored, if it can be picked up at all, and as long as everyone has their telly aerials pointed the right way (towards Sandy Heath) they won't pick up the local transmissions.
Rural broadband is the most obvious application of white space, and will probably be the first to be widely adopted. White space devices will have to consult an online database to check what frequencies are available locally, which is ideal for an access-point topology where the hub can check online and then tell client devices which frequency to use.
TTP, which already has a 5.5km link running at up to 4Mb/sec, reckons that such access will eventually be wholesaled nationally, with BT being the obvious provider. The recently approved 802.22 standard should fit such applications well, though both TTP and Cambridge Consultants (who have their own 6km link running at 2Mb/sec) reckon the database needs to be a two-way affair if it's going to work properly.
Ask and you shall receive
The current thinking for white space is that devices will submit their location and receive a list of the available channels in their area. But even at this stage, it is becoming obvious that static data will not be enough when a TV transmitter can stretch 60 miles or more depending on the weather, so both companies expect some nodes to report back to the database on local conditions, to the benefit of other white space users.
Quite how that works in practice remains to be seen: Ofcom has yet to decide who will run the UK database, and how, though an announcement is expected soon. The point here is that a one-way relationship isn't going to allow efficient use of the available radio frequencies.
Not that the database will ever be redundant – hidden transmitters cannot be sensed but could be interfered with – but using a database alone would require worst-case allocations, which would reduce the white space available, so a combination effort seems most practical.
Everyone involved in white space sees the technique as a key driver for cognitive radio technology. Cognitive devices are able to sense and avoid frequencies being used, and are still in their infancy. DECT phones will scan a handful of channels to avoid other DECT devices, and a few Wi-Fi routers can automatically select the most-empty channel, but that's detecting a common standard operating within specific bands. Cognitive white space devices will be expected to detect all kinds of use across very wide frequencies.
But if that can be achieved, and been proven in the white spaces, there seems little reason not to apply it elsewhere. Ofcom sincerely follows its remit to promote efficient use of radio spectrum, generally using auctions. This is based on the idea that whoever pays the most has greatest incentive to fill the bandwidth, but the regulator is always open to suggestions on better ways.
Sensing becomes even more important when one starts looking at some of the more innovative applications of the white spaces. Pop-up networks could provide connectivity across a festival or event site, providing decent bandwidth at a very low price. The 802.22 working group reckons it can fit 22Mb/sec over an 8MHz channel, which is probably optimistic, but there's nothing to stop channels being combined.