I’ve had several calls recently from clients asking about wireless mesh networks. Those that have been in the industry for a while probably remember the mesh network craze in the late 1990s. At that time large cities all over the country considered building WiFi mesh networks to try to bring broadband to everybody in their cities. Many cities deployed pilot systems, but in the end, the technology never panned out. The technology had the squirrely problems often associated with wireless technology and never delivered the bandwidth that the manufacturers promised.
Apparently, the technology is back. I went to the web for a quick investigation, and sure enough there are carrier-class outdoor mesh radios available from a number of different manufacturers. In case you aren’t familiar with the concept of a mesh network, it’s a network comprised of multiple radios, each of which connects to multiple other radios. Most mesh networks are dynamically linked, meaning that the radios work autonomously to find the most efficient routing path for traffic within the mesh. The easiest way to understand this is with this diagram from Cisco, which has been manufacturing mesh network gear for many years. In this diagram each radio interconnects with neighboring radios.
The biggest flaw in the technology two decades ago was that the mesh networks didn’t scale well. This was for two reasons. First, by definition, a wireless link loses half of its bandwidth with every hop to another radio. Mesh networks with too many hops don’t deliver very much bandwidth to the most remote nodes in the network.
Large mesh network also developed an unexpected problem. One of the characteristics of a mesh network is that the radios constantly coordinate with each other. If a given node is temporarily overloaded with a big bandwidth demand from an end user, the network dynamically routes other traffic around the bottleneck. Unfortunately, it turned out that in large networks the radios spent a majority of the bandwidth communicating with each other at the expense of the bandwidth left for end users. As mesh network grew in size the amount of bandwidth throughput decreased significantly. Technicians determined that this excess internode chatter could be reduced by limiting the number of nodes that any radio could communicate with, but in doing so the network was no longer a real mesh.
The other big problem in the 1990s is that the networks were deployed as outdoor radios, meaning that very little bandwidth actually made it into homes. I remember working one day at a client where I could see a nearby mesh radio through a window. As long as I sat where I had a direct line of sight to the radio I could use the WiFi, but if I moved to another part of the room the signal completely died. Broadcasting WiFi with outside radios is an inefficient way to provide bandwidth indoors.
Those inherent problems are still an issue today. There is no way to eliminate the issue of the bandwidth decreasing with each hop. However, the difference from today and the 19990s is that we can feed a mesh network with gigabits of broadband instead of with a few T1s. To some degree, that means that we can overpower the system so that at least some bandwidth makes it to the furthest nodes in the network.
One of the other weaknesses of a mesh network is that most networks use WiFi spectrum. Practically every wired home uses WiFi today to move bandwidth around the house. Superimposing a mesh WiFi network in a neighborhood means a lot more high-power WiFi sources to cause interference with every other WiFi device. Anybody who has ever tried to maintain a WiFi signal in a crowded business hotel understands the issues with WiFi interference.
Even with those limitations, I can see some great uses for a mesh network. The vendors are pushing the technology as a way to bring bandwidth more easily to outdoor spaces like parks. There is a brand of outdoor mesh devices being marketed as a way to spread WiFi around a farmhouse to the outdoor buildings. While nobody seems to be marketing the idea yet, a mesh network might be a good way to spread WiFi signals to fields and pastures to track the small bandwidth sensors being used to collect data from fields and herds.
What my clients really wanted to know is if a mesh network could be used to provide residential broadband. There might be situations where this makes sense. Rather than trying to beam the bandwidth from outside hotspots, each radio could feed a wire into a home. But mesh networks still have the same inherent problems as in the past and in most cases other solutions can probably produce faster and more consistent bandwidth. As a consultant I always have an open mind, but having seen the technology crash and burn once before I’d want to see this in practice before buying the resurgence of the technology again.