Champaign-Urbana Community Wireless Network releases first-generation mesh/cloud software, seeks input and development: I spoke with Sascha Meinrath, one of the folks leading the CUWiN project, about the scope of the project, their goals for outside participation, and his recent trip to Amsterdam to meet with a group designing documentation on wireless networks for developing nations.
The CUWiN project wants to allow self-forming, noncentralized, mesh-based Wi-Fi networks using standard, old PCs with no configuration. Slightly more advanced units could be ruggedized boxes using Compact Flash, but the basic unit would be a 486 or later PC with a bootable CD-ROM or bootable floppy that bootstraps a CD-ROM. Once booted, a unit finds other similar units without any other configuration or control and forms a mesh.
"We've been developing software now since about 2000, and our idea is to build software that is super user friendly, super easy for someone who doesn't understand the nuances of the technology or community wireless networking to set up their own system," said Meinrath. It's an attempt to enable community networking to spread beyond the folks who are self-starters.
To test their current software, they put together a bunch of old Pentium 133-based system with off-the-shelf Wi-Fi gear, burned CD-ROMs, booted the boxes and watched the mesh network form within five minutes.
However, the current generation of software "won't scale well: there's no route prioritization, and there's this problem of the hidden node problem," he said. (In a hub-and-spoke network, hidden nodes can see the hub not other spokes and can disrupt other network traffic by improperly sending at times when other nodes are transmitting resulting in interference and back-off behavior that reduces network performance. Mesh avoids some hub and spoke problems, but can effectively move the hidden node problem to any mesh point that has some connected nodes that can hear each other and some that cannot.)
CUWiN is design a system to prioritize routes among mesh nodes based on MIT Roofnet, and are looking into the Hazy Sighted Link State (HSLS) routing issue. HSLS uses packet economics: more dropped packets in a given route de-emphasizes it shunting more traffic to more successful routes. (Read more about this in CUWiN's FAQ.)
The software release by CUWiN of a CD-ROM image containing bootable node software along with the developer's resource (distributed under a BSD license with plans to move to a GPL license) is part of their effort to bring in more programming aid on the project. "We're relying on the open source community to provide us with feedback and ideas," Meinrath said. "A lot of our inspiration definitely comes from other wireless groups."
Meinrath noted that other groups are working on similar but not totally related problems. The HSLS issue is one that he believes no one has tackled directly and the group knows is central to providing the decentralized, non-hierarchical, dynamically prioritized system they want to offer.
Interested developers can download the CD-ROM image or contact CUWiN directly to get source code for the bootable project and the Compact Flash version.
On a somewhat related front, Meinrath recently returned from the first meeting in Amsterdam of The Tactical Technology Collective, which works with the Open Society Initiative and the Soros Foundation Network. The group's goal, Meinrath said, is to put together a resource guide for development nations that has all of the components for building a wireless network: regulation, configuration, installation, and other details.
Participants came from around the world, including Denmark, the U.S., Senegal, Indonesia, Canada, and London. The members aren't just from developed nations, but include people out in the field in community wireless in developing countries.
"The focus is on training manuals and on those sorts of resources rather than implementation, which is good and bad -- which is good and incomplete," Meinrath said. "One of the problems that we see a lot is that people develop really cool hardware that's far beyond the means of people on the ground where this equipment is supposed to be used." The group hopes to have the initial documentation complete in six months.
The Tactical Tech group's social goal is partly the motivation for the CUWiN project as well: to develop technology that relies on its intelligence rather than the cost of the components, making the use of several generations-old computer technology feasible.
CUWiN's current disk image is about 30 Mb, and they have a beta that requires just 13 Mb. Their ultimate goal is to reach 2 to 4 Mb in size so that it can be flashed (written into erasable programmable memory) onto commodity units with those limits.
The current generation of Compact Flash CUWiN software has solved a vexing problem: they can now flash upgrade units by connecting to them wirelessly without swapping out the Compact Flash card.
Meinrath said fundamentally, "We're most interested in how to build a system in which anyone, not just a techie, can set up a mesh with as few as two and as many as 1,000 nodes."
He offered some recommended reading on both the education site at the Free Press's Wi-Fi section (which includes excellent graphics about different community network configurations), and the CUWiN grant from the OSI.