Features

Become a wireless ISP: for three hundred pounds

While the learned are laughing at Negroponte's fantastic "futuristic" vision of a mesh of interconnected wireless LANs "like lilypads which you hop from one to another" a UK company has produced Mesh wireless technology which you can buy and install, today, for under £300.

Fancy setting up as a rival to BT Openworld? Even in a remote village? Easy: buy a Locustworld MeshBox; half the price of a home PC. You're in business.

The software is the key to Locustworld. Written by text-message pioneer Jon Anderson, it configures a group of wireless access points into a coherent "mesh" and connects them to any broadband Internet node available.

Most experts regard the mesh approach as hugely complex, because of the effort needed to set up the mesh. The system used to be known as a "parasitic network" - although the fashionable term these days is "symbiotic" - the idea is that you turn a group of wireless nodes loose, and tell them to introduce themselves to each other.

Then you set up routes through the mesh. It can be fiendishly complex, but Locustworld's mesh does this for you. You just buy the node from them: the current model is £250 plus VAT.

The last legal obstacle, according to founder Richard Lander, was the decision by Oftel, allowing people to share their broadband with up to 20 others.

The excitement in the UK hasn't been quite as high as it was in the US, but even there, it seems only "nerds" really picked up on it - probably due to an article by Anderson which was flagged on Slash-Dot in December.

It should have hit the headlines big time, since it allows a street to share all their broadband nodes, at a huge cost saving. It would allow a vicar in a small village to hire a leased line, and share the costs with all his parishioners - without any technical expertise.

In the article, Anderson describes the cheapest way of setting up a mesh node; by installingthe software, from a CD onto any PC. It's only 32 megabytes of code, and it is also available in a Flash memory card. The response, however, was a typical Linux community one: "Why doesn't he use this open source utility rather than that one?" and "I bet I could buy a fanless PC for less than that, and set it up" and "There's a better motherboard available from Taiwan."

Now, Locustworld has released the full Meshbox: a standalone 500 MHz (fanless) PC, suitable for installation in any living room next to the audio equipment.

Its simplest form is with a single antenna, which works on WiFi (802.11b) standards anywhere in the world, and provides shared access to the PC, but also looks for other Meshbox installations in the neighbourhood. There's a second option; an additional, long-range antenna, which you can mount on the roof of your house, to pick up signals from other Meshboxes further away - across the village, perhaps.

And of course, since it is actually just a PC, the Meshbox can be used as one. You can plug in a monitor and surf the web or do email. It does network address translation for sub-net members, and issues DHCP leases. In short, you plug it in, and it works, more or less.

The impressive part is the ability to combine several broadband feeds. This is an option (it doesn't happen out of the box, but it's built in) - if there are 20 houses in the Mesh, but only four of them have broadband, all users can share all of those lines. It does add to network traffic; there's an estimated 10% increase in wireless load, says Lander.

That's the easy part. And (without doing a full review) it does seem that this has been within the capacity of quite inexperienced computer users to set up. However, there are going to be some controversial areas in the Locustworld experiment. The cheekiest move was the setting up of an IP address numbering authority, WIANA, or The Wireless Internet Assigned Numbers Authority.

"If you have two meshes expanding towards each other, and each assigns IP numbers without regard to what else is going on in the world, you will end up with a numbers clash at the borders," said Lander. "Our WIANA makes sure that no two meshes ever conflict."

The achievement is pretty startling, considering the number of people who have tried to do a wireless mesh technology, and discovered it to be very expensive.

Perhaps the most startling contrast is with Mesh Networks in the US. This is a heavily funded R&D-based organisation, which has developed its own proprietary 2.4 GHz wireless technology (requiring special licensing) and which charges in the region of $2,000 for a client access card (a WiFi card costs under $100 these days) and the node computer fetches $4,000. But the irony is that in the same edition of Wired magazine in which Negroponte recently speculated about the feasibility of "lily-pad" networks, there was a brief mention of Locustworld ... which excited no comment at all.

So far, scepticism has been the main reason for caution.

"Parasitic networks are pretty damn complex," remarked one expert inside BT's research department. "We're close to releasing some technology, but there's a lot of testing to go through."

Another example of "it's complex" was the approach to linking sites by wireless taken by Radiant Networks, which uses a very high-bandwidth wireless technology to distribute broadband to a number of antennae. The masts are tall and expensive; the wireless technology is licensed, not 2.4 GHz, and two trials have been set up under the watchful eye of the DTI, which has yet to report whether it's viable as a way of providing rural broadband.

So far, Locustworld sales have been encouragingly global, but not yet large-scale in implementation. That's hardly astonishing: the network hasn't been available until late last year, and the announcement of a self-contained mesh node for half the price of a PC was made just before the end-of-year holiday season, and got missed by most potential readers.

But there is something to see. There's a pioneer network at Kingsbridge in Devon, and there are trial sites in Macedonia, India, USA, the Netherlands, and of course, in several UK counties.

The implications if this technology starts to get mass market availability are scary. The Guardian guessed that the biggest sufferer would be third-generation phone networks, which are counting on high-data usage, not voice, for their revenues; but a universal WiFi mesh in all inhabited areas would deprive them of the bulk of their income.

On the face of it, the mesh network could grow instantly. You can download the software itself from the Locustworld site and set it up on any machine capable of booting from compact-flash or CD.

But experience teaches that mass market products need mass market support. That means distributors, resellers, installers, advertising, promotion and branding.

None of that, clearly, is yet available from Locustworld itself, nor is it feasible to predict that it will be inside the next two years. Not only is Locustworld designed to be a grass-roots, subversive structure for personal "empowerment" but it also undermines the credibility of any well-funded commercial enterprise; on the face of it, how could you supply such products with a margin, if people can read about it on your advertising, try it out at your expense, and then build it themselves?

At some stage, some enterprising company will get the point, smell the toast, and start packaging a solution. It may be a company like Richard Nuttall's Invisible Networks which is currently using rather more expensive techniques to bring broadband to rural areas. It may be someone much bigger, like Telewest or BT Openworld or it may even be a company like Hutchison 3G which first sees the writing on the wall and decides that if anybody is going to pinch its profits, it will be Hutchison itself.

But it won't happen overnight.

The bad news is that the first response is most likely to be a rash of rivals, energetic and opinionated nerds who think they can improve on the Locustworld model. And, of course, some of them will like the idea of WIANA, but think that IPv6 addressing will be a superior way of resolving it. Lander doubts this: "It's just another way of giving big ISPs blocks of addresses, it won't solve this problem." But that won't stop people from producing meshes that don't interwork with Locustworld, and getting largescale adoption by whole communities.

By the time a genuinely global corporation with the power to co-ordinate a standard solution gets the scent of profit, there could well be a Tower of Babel of other networks, things like Sputnik or Boingo who focus far more on the enterprise markets, where Locustworld will have trouble reaching, and all clamoring to be accepted as the best.

But it is a step in the right direction, isn't it?