Wednesday, February 22, 2006

Wind turbines for mesh networks

Late last year Nature carried a news snippet about mini-wind turbines for powering wireless networks, which cited a paper in Applied Physics Letters by Shashank Priya which talked about using piezoelectric materials and mini-turbines for distributed power generation. I actually meant to post something about it at the time, but the link has sat in my edit queue since November.

However Engadget's post about the micro-windmill for charging your mobile phone, developed by an Indian graduate student, started me thinking about wind turbines and mesh-networks again.

The interesting thing about mesh-networks is that they don't really operate in the same way as more traditional network designs, they're often compared to cellular networks but the comparison is flawed. In mesh networks each network node participates in the routing process, relaying packets intended for other nodes. This isn't the case in a cellular network where routing takes place only over the backbone network connecting the cell base stations, rather than between individual handsets although I've talked about how to use Bluetooth to do mesh networks between handsets in the past.

Of course the really different thing about a mesh network when compared to traditional network designs is that mesh-networks are ad-hoc in nature, and are fully self-configurable and self-healing. After distributing the nodes over a widespread area (perhaps by parachute drop?) the nodes themselves discover their nearest neighbours, and discover routes to other nodes in the network independently. If an existing link fails then network automatically reconfigures itself to compensate. Funnily enough this aspect of mesh-networks means that a lot of work has been done on them by the military, whose network nodes (e.g. tanks, soldiers) tend to get shot out from under them a lot. Although after a long time as specialist equipment, gear to do mesh-networks is starting to become mainstream although it's not exactly a cheap solution at the moment.

However in the past mesh-networks have tended to rely on batteries, which eventually have to be replaced, or mains power, which in a lot of situations defeats the point. This probably comes from the military background of most of the work done on them, the military usually aren't that interested in long term operations and aren't really that concerned about cost. Mesh-networks are perfect to networking large sparesely inhabited regions, think third world countries, or to serve as the network backbone for a ubiquitous computing environment in a more developed and urban environment. The holy-grail of mesh-networks would be to make the node itself cheap enough to be thrown away. You air drop the nodes out of a transport plane and they're basically self-sufficient, and when they do finally fail, they're cheap enough to be replaced without an expensive operation to retrieve the first one. For that, you need distributed power generation.

I've talked about distributed power generation before, and of course wind power isn't the only solution to the problem. With spiralling energy costs centrally provided power might eventually become a thing of the past as more people want to take control into their own hands.

Technology is giving us a choice, we can follow the path of central control, or that of decentralisation. More and more, technology is giving governments or large corporations the power to federate information and monitor its flow. Alternatively, it's also giving us the power to decentralise. Things that before were only provided centrally because of the prohibitive cost, such as power and networking, can now be distributed, and done entirely on a peer-to-peer basis. We face a cross roads, and the technology itself won't tell which way to turn.