Craig Mathias, Computerworld

I'm a big fan of metro-scale Wi-Fi networks, which are now being deployed around the world. Some of these projects are very exciting. For example, the entire country of Singapore is busy unwiring, using three different operators to cover virtually all citizens and businesses.

However, while these projects are exciting, metro-scale Wi-Fi networks can't provide the same coverage as wide-area wireless technologies such as cellular 3G and mobile WiMax. One reason for that is that mesh infrastructure such as access points or nodes used for metro-scale Wi-Fi must be located in fixed locations. Sure, we can add more nodes and more radios per node, but given the technology, and because the nodes need power, they can't move around. As a result, the current generation of metro-scale Wi-Fi mesh networks are appropriate for high-density, high-demand areas, whereas 3G, WiMax and similar technologies are particularly useful when wider coverage is essential.

However, imagine mesh nodes that aren't connected to a specific location. This mesh of the future will have two key properties. First, the mesh - even potentially a very large mesh - can be deployed very rapidly. This capability would be ideal for public safety, homeland security and many more applications, such as an enterprise needing to cover an area on a quick-deploy or temporary basis. There's no need to install the nodes - just set them up and turn them on.

A second property of this type of technology is that the nodes can actually move during operation, which makes things a lot more interesting. This means we're talking about a fully mobile infrastructure, not just mobile clients. This opens the possibility for a truly dynamic mesh architecture, one that can grow organically in any large-scale venue, whether it be in a public space or in an enterprise.

A vehicular mesh?

One concept we've been exploring is mesh nodes that are deployed on vehicles that are commonly on the move, such as buses, taxis, police cars and fire trucks. These moving nodes would form the core infrastructure of the mesh. Client nodes would associate with these mesh points and re-associate with others as a given node roams out of range. Backhaul can be provided via relaying through the mesh, using other wireless technologies like 3G and WiMax and/or a few fixed points that provide a bridge to wired capacity.

This approach needs highly complex mesh-routing algorithms, but significant work is being done in this area. One company that has staked out this space is PacketHop, which was spun out of SRI International a few years ago. PacketHop is currently specialising in dynamic mesh applications for public safety and first responders and has reported some interesting test and production deployments based on the concept.

One of these, in Long Beach, used PacketHop's products for a simulated homeland security exercise at Long Beach Airport earlier this year. This approach was particularly valuable in this case because it provided a single, unified, multimedia network that could be set up quickly and that could dynamically and quickly adapt to any situation, no matter how the focus of activity moved geographically.

This same technique could be applied to commercial and enterprise applications as well. Imagine, for example, every car being a dynamic mesh node, and the amount of capacity that this technology would create on and near almost every road. Imagine being able to cover a corporate campus, providing access to vehicles and buildings alike.

Eventually, fixed and dynamic meshes will merge, but that's a topic for another column. In the meantime, keep your eye on this concept - it's going to become very important over the next few years.