FreeWANs - Part I

Originally published here.

After reading William Keeley's article on setting up a FreeWAN cell, I was intrigued by the possibilities of communicating over a network of FreeWAN cells. Would it be possile to connect to one outpost cell and use it to "hop" to another? Could a network of FreeWAN cells become searchable? Could this type of network be shut down in an instant by some nasty government authority?

After a little bit of digging and it seems the answers come in at: Yes, Yes and Not Really. Here's the low down.

Mesh Networks

Mesh networking is entirly based on small, self organising nodes working to create a decentralised wireless network. Take Gnutella and imagine it existing in the physical world - that's mesh networking.

The US Department of Defense has been mucking around in this field of communications since the early 1970s and is currently working on a fully fledged mesh network called the Joint Tactical Radio System for deployment in the next few years.

And what exactly do these nodes for mesh networks look like?

Obviously, that's just the controller chip, but basically, you can squeeze a outdoor-network-ready mote node down to the size of a USB flash drive. Due to the extremely small size, power usage can be a problem but a three-year working life on 2xAA batteries ain't too bad now, is it?

Some designs currently in development are even able to supplement their batteries by drawing power from vibrations in the environment around them, like those watches that get power from the movement of your wrist. Remembering where you left your mote three years ago may be a problem, but luckily there are many GPS options available to help you locate the little buggers for battery changes.

Many of these commercial grade motes run on the 802.15.4 wireless network band, with speeds of up to 250 kbps. Intel has actually been looking at mesh networks quite seriously, and have been conducting some very interesting research. For instance, Intel's EcoSense project has been looking at "how to network large numbers of inexpensive wireless sensor nodes while maintaining a high level of network performance." They came up with some cool ideas too.

Heterogeneous Networks

In another example of a multi-billion dollar company "discovering" things that were developed ages ago by Open Source programmers, Intel has worked out what an Ultrapeer is.

By taking a mesh network of low power motes and overlaying a series of high powered Intel XScale based nodes, the performance of the network can be significantly increased. The XScale mote also solves the problem of getting data from an 802.15.4 network onto the 802.11 part of the spectrum.

Experiments with the XScale overlay show that mesh networks can be greatly improved by the presence of the Ultrapeer-like nodes (wow!), and by adding even bigger hardware to the network, like a regular home router, the reliability increses even more (who’d have thought it?).

And there we have a bridge to link two FreeWAN cells.

Creating Searchable Networks

Intel takes a leaf from the FLOSS book. Again. Where there's TinyOS, the open source operating system which has become a standard for building mote networks, Intel came up with TinyDB, a query processing system.

"The ultimate goal of TinyDB is to allow people to query sensor networks without having to program them," says Mark Yarvis, the principal investigator for Intel's Heterogeneous Sensor Networking project. "Essentially, TinyDB imposes a database model on top of the sensor networks. Once the model is in place, users can do sensing tasks based on simply posing queries, similar to querying a standard database."

Intel is trying to work out how to optimize a mesh network using TinyDB. "Even in a network of completely homogenous hardware, some nodes may be underutilized. That makes them good places to cache data to handle complex group queries or similar tasks." Any sort of search engine would use the resources of these underutilized nodes, so in a network made up of mixed FreeWAN cells and smaller motes, chances are it will still be a regular PC with a few clock cycles to spare. Kind of like Ultrapeers really, except the resource barrier will be lower with TinyDB code running the show.

Unfortunately, the best application Intel can up with for this type of network is in theme parks. "One potential use is monitoring the quality of water in tanks", according to Intel, but they do make up for the mundane thinking.

"If you're at the park with your kids all day, you could use the (mote accessible) Internet to stay in touch with friends and the office as well. And the kids could download and store information for a school paper. In general, it would provide a way to stay connected," says Lakshman Krishnamurthy, Principal Investigator the Intel EcoSense project.

W00t!

Well, now we know what's possible. Next, we'll look at some of the problems you might face building your own alternative internet.