I can see why it took two years to develop a mesh networking topology for Bluetooth. Basically, the members of the Bluetooth Special Interest Group are shoehorning the low power personal area network into an entirely new use case. The SIG and its members want to tailor Bluetooth, known for fitness trackers and headsets, into the industrial and enterprise IoT markets.
As such, it’s getting new capabilities, a new network topology and an entire second layer of software that someone is going to have to manage. There are even some aspects that the SIG hasn’t solved yet. Read on for all the deets.
First up is the network topology. That’s just a fancy way of saying how the network is organized. Instead of a traditional point-to-point network, Bluetooth now has mesh capabilities where all points can talk to each other. There is an overarching trend toward mesh networks as we connect more and more nodes to the network. Wi-Fi mesh networks are all the rage in homes and Thread, the newest wireless protocol, is a mesh network.
There are two types of mesh networks. The first is a routed mesh, where individual devices in the network have paths of conversation. ZigBee, a popular wireless protocol for lighting, is a routed mesh. The second type is a flooding mesh, where every device on the network can shout its messages to everything else. Some of the original Bluetooth mesh efforts behaved this way, which led to challenges adding too many devices on the network. It’s also less power efficient.
The Bluetooth SIG has split the difference, deciding on a managed flood for its network. This means that certain devices can pass along messages but not every device can. In many cases, those devices capable of passing along a message will be wired into power, such as a light bulb.
To go along with this new complexity, the SIG has also developed mesh models for different use cases. The two use cases that appear to have mesh models so far are lighting and sensors. In lighting, the emphasis is on turning everything on or off at once, scalability and covering the necessary area. In sensors, the focus is on letting battery-powered devices sleep for as long as they can, and then ensuring that when they wake up they can send and receive messages that have been waiting for them. Power conservation is key.
These mesh models sit on top of the radio layer, so this is going to be something that the builders of connected devices, hubs and perhaps even chip vendors will have to manage. The idea is that when a consumer or end user provisions the device they will set up the profile of the device so it can follow the role in the right model. That sounds complicated, so we’ll have to see how this works out in practice. A great UI will make all the difference.
We shouldn’t have to wait too long to see products using Bluetooth mesh. Because of the long-gestation time of this standard, devices are going to hit the market quickly. The mesh software should work on any Bluetooth Low Energy radios that are 4.0 or later. The original audio-streaming Bluetooth radios need not apply. Many chip vendors have anticipated the needs of Bluetooth mesh and have been building chips that can be updated over the air. This means that products will soon get an update that turns them from Bluetooth’s original flooded routing approach to the new, managed flood approach.
There are some other elements worth noting here. Unlike the Thread wireless protocol, Bluetooth mesh networks won’t have the ability to let devices talk to the Internet through IPv6. This means a Bluetooth mesh won’t span more than one IP address, according to Ken Kolderup, the VP of marketing for the Bluetooth SIG. In practical terms, your Bluetooth mesh will be limited to a specific building or area.
And because the SIG didn’t want to face some of the interoperability problems other wireless protocols dealt with, when building purpose-built software layers on top of a radio standard the current mesh models for lighting and sensors are frozen. You can’t change that going forward. Kolderup said this was a way to focus on keeping interoperability and that the models are very flexible. The SIG wants to ensure that screwing in a light bulb to an existing Bluetooth mesh twenty years from now works, and because, as Kolderup says, “I don’t know what a light switch is going to be in 20 years.”
The idea that Bluetooth mesh networks will be around in 20 years shows how far the SIG has gone to try to build something that industrial customers, enterprise customers and even home builders can put in their products. And while I believe that Thread has more modern capabilities as a wireless mesh networking protocol since it was designed from the ground up in the last few years, Bluetooth has the advantage of familiarity and that it’s on everyone’s phones.
I doubt we’ll be controlling many of these networks with our phones, but the economies of scale associated with selling hundreds of millions of devices certainly give it a cost and credibility advantage. Either way, mesh networks are here to stay. The internet of things demands them.