The beauty of mesh networking

The first time I heard about mesh networking, I was at a 50-person conference about how to keep the Internet from being turned into a clone of the phone system. One of the attendees—all of whom knew more about the topic than I did—talked about a geekfest he’d been to in Berlin where about 50 little mesh networking cubes were scattered about the city. I pictured them lighting up, one by one, as they found one another and created a network out of nowhere. The cubes themselves were cheaper than wifi routers at the time (and still are). The conference attendees were excited by the experiment, I think perhaps by the beauty of the networking model as much as by the realistic prospects.

The model has a lot of symbolism going for it. Wifi routers create a one-to-many relationship, not unlike a TV broadcast system, with the important difference that wifi routers obviously are two-way. Nevertheless, coverage of an area relies on someone providing access to those within some radius of radio waves. If the Smiths down the street turn off their wifi when they go on vacation, that portion of the neighborhood goes dark.

Mesh networks work differently. Each mesh node is also a router. If it’s within radio signal of other nodes, it automatically creates connections. Each node routes messages to the next hop, and if the Smiths have turned off their mesh node, your message will find the next available node to get it closer to its destination. Of course, if the Smiths have the only mesh node within reach, your message isn’t going anywhere.

But that’s part of the beauty of mesh networks. They exhibit what Internet architect David Reed calls "cooperative gain": The more who participate, the more bandwidth. That’s the counter-intuitive opposite of most other networks: The more people who drive on a road, the less available capacity there is, until at last you all try to get across a bridge at the same time and come to a complete standstill. In normal networks, you divide capacity by usage. In mesh networks, you multiply (so to speak).

Robin Chase, the founder of Zipcar, gives a particularly powerful analogy. One strategy for getting across a broad, narrow stream, she says, is to build a bridge that connects the major road on either side. That works, except that when it’s crowded, you run out of capacity and cars have to wait, and when it’s empty you’ve built excess capacity that’s going to waste. The alternative, says Chase, is to have everyone who comes throw a flat rock into the stream, forming pathways that let you hop to the other side. The more people who come, the more rocks, the more pathways, the more capacity. Cooperative gain.

There are limits, of course. Chase points out that if you’re trying to get your message from Sheboygan to Kabul, you’re going to have to use the highway. Or, to de-analogize, if there are a hundred mesh boxes in your community, but only the Mendocinos’ is connected to a cable router or DSL line, those hundred mesh boxes are going to be slicing up the Mendocinos’ bandwidth. That bandwidth the Mendocinos offer is not going to magically grow. Of course, if a quarter of those hundred mesh boxes are connected to the Internet, the hundred mesh nodes may be able to take better advantage of the excess capacity of the 25 Internet pipes available to them.

But the really exciting thing about mesh networks is how they open up the possibilities for local networks that do not have to go out on the Internet. Messages can be passed from one node to another without the Internet ever being bothered or alerted. Imagine a neighborhood where mesh networking is as common as Internet access. In fact, because you can set up a mesh node without having to pay an Internet access provider, mesh could be more widespread than Net access. People in the community could talk, set up calendars, create pages, coordinate rides, send out missing cat alerts and play Texas Hold ‘em. A local mesh network—and all mesh networks are local—is tied to the geography it covers. This creates a new platform that’s potentially as important as Facebook, because geography still counts when it comes to sociality. And education. And governance.

There have been concerns about mesh networking’s practicality. In particular, can it scale? The answer seems to be yes, based on implementations in cities in Europe. Vienna, for example, has a mesh network that extends 30 kilometers. Boston is on the verge of prototyping mesh networks in two areas of the city, one affluent and one not so much. I suspect the element of risk is not the technology, for that seems to be working well enough, but getting the right set of apps in place to entice people to reach out and mesh someone.

Thus the glory of mesh. It forms itself bottom up. It creates more capacity. It rewards cooperation. It reminds us of the primary importance of the truly local. It is a low-lying cloud that shapes itself to the terrain. It is beautiful.