Everybody Runs On Bluetooth Now

Back in 2018 I commented on LinkedIn that [short range, low power] wireless war was over and Bluetooth won. At that time this true statement of fact could not have been backed by hard data points. So the technology supremacy debate has continued and is still active. But to me that was obvious then. Partially based on the fundamental physical characteristics of Bluetooth LE, such as speed, power consumption, spectral efficiency - each being superior to any competing technology. Partially based on the allocation of frequency bands, where, again, Bluetooth offers much more and is globally unified. Partially due to direct connectivity to phones and tablets that Bluetooth allows for, creating the simplest, most streamlined user experience. And partially because Bluetooth is all on fire, when long term evolution is considered.

People who are not inside this development tend to judge the situation by looking in a rear view mirror. "Bluetooth has never been used to control lighting systems" - I was told back in 2014/2015, when we were laying down the foundation for Bluetooth mesh. Now it does, at commercial / industrial scale with success proven by case studies (like this) and hero products (like this and this).

The rise of programmable radio SoCs has opened the door for players like Thread to use Bluetooth. Why would they do it? Well definitely Bluetooth must be solving some of their problems, with the UI (commissioning) being probably the most intense. Then there is this concept of wireless services like beacons and asset tracking, with both applications being already dominated by Bluetooth. So the multi-protocol hybrid has become increasingly popular.

Signify decided to take advantage of the same hybrid opportunity, to launch the Bluetooth - based Philips Hue lights. Finally we do not need a hub for Hue. True. But internally Hue still run a combo Bluetooth LE + ZigBee stack.

Both Thread and Hue are like Toyota Prius being a hybrid evolution step on the way from the combustion era of Benz to the electric era or Tesla. It is clear all-electric is the way to go in the long run, it is just that not everyone (for many reasons, like legacy support) can get there in one step.

But the problem with hybrids is the complexity. Priuses need to have both their combustion engines and batteries manufactured and serviced. In the end the combustion engines will go and the future will be radically simplified.

In the low power wireless domain there is one fundamental problem with hybrids. While the chip vendors claim the multi-protocol capability and on-the-fly switching between, say 802.15.4 and Bluetooth LE, this has significant limitations. A chip cannot listen for both at the same time - because it has just one radio, capable of tuning to one particular frequency. And if it is in the middle of receiving a Thread packet, it is deaf to Bluetooth transmissions. And vice versa. What does that mean? Sub-optimal performance. While this may not matter for small scale home systems, this is fundamental for commercial and industrial applications. Like Prius, such hybrid may be OK for personal use. But for large scale / high performance this is not an acceptable architecture. In a mesh network the nodes simply cannot afford losing any incoming messages just because they have been busy with something else. If you lose 20% of incoming messages per "hop", the probability of that message reaching a destination over 10 hops is just 10% (0.8 to the power of 10). Meaning 9 out of 10 messages would be lost.

This performance issue and the fact that running two protocol stacks on a chip (Bluetooth and something else) adds significantly to the overall complexity (read: cost), I don't think the hybrids will have a long successful life. From a practical standpoint, once Bluetooth is already on a chip, why use anything else? There would have to be some important reason to run Bluetooth and that second stack in a hybrid configuration.

One is definitely support for legacy products, which justifies why Hue is still using ZigBee.

Another one, being brought so often is support for IPv6 over a low power mesh network. Bluetooth mesh does not do that. Yet. On the other hand, based on the market feedback and industry reports, things like data reliability (92%), battery lifetime (92%), data security (87%), equipment cost (82%), network range (78%), service fee (60%), standards (45%) are more important purchasing considerations than IP addressability, which is 27% (based on 2019 ON World's survey).

The bottom line is, and this has been proven by the recent announcements by Thread and Signify, Bluetooth is here to stay. And it is the technology that most other technologies will converge to. It was my bet back in 2012 when I started exploring the potential for using Bluetooth in device networks. So 2019 is just a datapoint on that projected trajectory. As was 2017 when Bluetooth mesh went gold. And there are similar points on that trajectory ahead of us. No, it has not been a smooth sailing. It has been the most fierce push by the SIG members community and product companies to make that technology successful. What is great now is that we all have full confidence in this trajectory.