Your browser is not supported. Please update it.

23 August 2018

Apple throws weight behind Thread, but ZigBee claims 500m chips

Apple has very quietly signed up to the Thread Group, the body pushing the 802.15.4-based low power mesh personal area network (PAN) protocol. It comes in the same week as Thread’s rival, and sometimes good friend, Zigbee passed 500m sales, with projections to hit 3.8bn by 2023.

However, the Zigbee Alliance is claiming that it will account for 85% of 802.15.4 shipments, which leaves very little room for Thread.

This could be huge news for Thread, especially if Apple incorporates it into HomeKit. However, both Apple and Thread are well behind Rethink’s expectations for smart home progress – although in their defense, the entire smart home sector has underperformed so far, especially in the smartphone market.

Apple only really has two devices that make sense as homes for Thread radios – the new HomePod smart speakers (Apple’s is not particularly smart compared to Amazon’s and Google’s), and the Apple TV. Given the generally slow pace of Thread device certification, we don’t expect Apple to be announcing much in the way of Thread compatibility over the next few months, and we would estimate that its involvement in Thread will be mostly from its HomeKit software framework – which governs how smart home devices interact with its iOS and Mac OS platforms.

Adding Thread would also be a notable departure for Apple, which until now has stuck solely to WiFi and Bluetooth, despite the option of the long-tenured Z-Wave and Zigbee protocols. Bluetooth 5.0 has introduced its own mesh networking feature, Bluetooth Mesh, which might prove a useful tool for the likes of Apple, and Bluetooth’s main benefit over the rivals is that it directly links smartphones and smart home devices – although a bridge or gateway would admittedly solve that particular problem.

Apple has recently exited the WiFi router market, and so it seems unlikely that it’s going to bother launching a new AirPort with Thread support. However, Apple should definitely consider the mileage it would get from its own brand of smart home devices, such as cameras, environmental sensors, and perhaps even a thermostat. It has a pretty dedicated group of core enthusiasts, as well as a strong brand that could go toe-to-toe with the likes of Nest – and what better way to ensnare people inside its ecosystem than to turn their house into a churn-cutter itself.

But again, to reiterate, Apple has been very slow in pushing HomeKit out. Its recent iOS 12 update only made minor tweaks to the Home app, and HomeKit has only just added support for third-party remote controls, and still limits cameras to mains-powered only.

The new HomeKit Made For iPhone (MFi) developer kit won’t do much to smooth over the developers that are jaded by Apple’s shift from mandated hardware-based security, on specified approved chips, to a software-based approach. Apple does say that the new dev kit has removed a lot of the burden from developers, and claims that prototypes only take a week to create, with full releases possible in just three months.

If developers are able to hit those speeds, then perhaps Apple might be able to significantly ramp HomeKit adoption. So far, a trickle of devices have arrived on the market, which leaves Apple in something of a catch-22 – why bother investing resources in HomeKit if almost none of its customers are using it?

Well, if the Zigbee Alliance’s proclamation is true, then there will be a whole heap more smart home devices in the world, which might eventually change Apple’s mind. According to figures from ON World, some 500m Zigbee chipsets have been sold to date, and by 2023, that figure is expected to hit 3.8bn. That’s some 85% of the total 4.5bn IEEE 802.15.4 chips that the forecast predicts, the family that includes Wi-SUN, and of course, Thread.

Thread and Zigbee use the same 802.15.4 MAC layer, and to this end, that does mean that some chips can support both protocols, meaning that they can share the same PHY layer. However, above the MAC, they are pretty different. Thread uses 6LoWPAN for the networking layer, whereas Zigbee has its own IP-compatible implementation, and then Thread also uses the Zigbee application layer. As you can see, it’s currently a bit messy, and things would be a lot easier if they could just merge already and iron out their differences.

Back in 2016, Zigbee and Thread had moved closer. The following December, in the run-up to CES 2018, the pair announced that Zigbee’s Dotdot application layer had been successfully ported to Thread, and that this joint stack constituted the first interoperable IoT language running over an IP-based network. Notably, Dotdot would also work as an option for things like WiFi, Bluetooth, or NB-IoT, should you be that way inclined.

Essentially, Dotdot defines how devices connect to networks and how they speak to other devices once they have joined. As long as a device is using Dotdot, it should be able to communicate with all other Dotdot devices on a shared network. While the Zigbee Cluster Library manages this for Zigbee, Dotdot is providing a way to do the same for other protocols – which has obvious benefits when it comes to IoT environments, as it means that WiFi devices should be able to communicate natively with Thread or Zigbee units.

In theory, Dotdot should enable all those devices to stay connected to all other Dotdot units, preventing them from completely losing their network connections. For developers, it means not having to hire the skills to support multiple protocols or devices, and for those looking to build or sell ecosystems, picking Dotdot should mean that your gateways, thermostats, or lightbulbs will all play nicely together – again, in theory.

But now that Thread-advocate Nest has been rolled back inside Google, we might see something of a more united front from the smartphone platforms. Chiefly, if Thread becomes the option of choice for both Android and iOS, then we might see some unification from the two camps, settling on a common approach that then might help the collective market grow.

As for the rest of ON World’s numbers, it says that Zigbee accounts for over a third of the current smart home Wireless Sensor Network (WSN) chipset market. The Z-Wave Alliance recently announced its 100mth chip shipment, for some perspective, although it’s worth noting that a lot of Zigbee chips are being used in things like TV remote controls, and not really smart home devices. By 2023, the figures project that Zigbee will be hitting one billion annual shipments.

One company with an interest in the ramifications of Apple’s latest move is Qorvo, and Wireless Watch’s sister service, Rethink IoT, spoke to Cees Links, general manager of the firm’s Wireless Connectivity business unit about the issue.

Links was CEO of GreenPeak, a specialist in Zigbee chips that Qorvo acquired in 2016 for $118.7m, and was very involved in the development and commercialization of WiFi and the launch of the WiFi Alliance.

Links posited one explanation for the apparently reluctance of Thread and Zigbee to merger – that the likes of Apple and Google, now two of the largest drivers behind Thread, are perfectly happy having a closed application layer, as it lets them build their own ecosystems on top of Thread, in a more closed fashion than one would using Zigbee or WiFi. The walled garden approach lets them keep the money close.

As such, the fight is in the application layer, and Links points to a real world example of the problems here – his Netgear router with five gateways hanging off of it, which is required to run his smart home. He notes that the fight seems to have moved from the radio to the application layer, and that the main conflict driver now will be the collection of data, or rather, the means that Apple and Google will use to ensure they can still collect all that valuable data.

Links goes as far as saying that the growth of the smart home has been hampered by these big players, that it’s not a lack of awareness from device makers driving the slow uptake. Illustrating this is the fact that nearly every Zigbee Alliance member is also a member of the Thread Group, except for Apple and Google.

Bluetooth’s latest mesh launch could be kicking off a new radio war, said Links, although he notes that Bluetooth’s attempt to take on the 802.15.4 radios might go as well as Bluetooth’s attempt to tackle WiFi some 15 years ago. On this point, Links added that most Zigbee chips are able to do Bluetooth communications already, and that he thinks most Zigbee chips will be combo units, as the additional cost is close to zero, as it’s nearly all done in software.

The driver for such a move comes from the benefits that Bluetooth, a connectivity protocol, brings to pairing Zigbee, a networking protocol. Links said the ability to use Bluetooth to connect a device to a smartphone, in order to carry out pairing and provisioning, and then let the device switch into Zigbee for normal operation, was very useful – especially as a way of quickly adding a screen or keyboard to a Zigbee network.

That’s a use that Qorvo is especially interested in. Links believes Qorvo is unique in being to run both Zigbee and Bluetooth simultaneously, although he does note this limits the device to one incoming packet, as that packet’s preamble header is what defines the radio response.

The company reckons it has around a 50% market share in 802.15.4 chips, but it remains very focused on specific verticals – remote controls, lighting and retail labeling, being most prominent. Links notes that it is easier to sell larger deals in these verticals, in thems of chips range, than to try and sell smaller deals across wider horizontals.

As for wider Zigbee options, Links says Qorvo is not interested in using the 900 MHz options for Zigbee, as regional differences in the precise band allocations have proven very off-putting for device makers. Links likened it to the problem of different power cords for early mobile phones all over again.

He added that one of the biggest annoyances with the persistent idea of running Zigbee in the 900 MHz range is the fact that you would either need a very large cheap antenna or a very expensive optimized antenna to make it work. This rules it out from most smart home devices.

However, Links said that in an age where multiple WiFi access points inside a home are becoming more common, the need for low power mesh protocols like Zigbee and Thread may be coming to a close – after all, you don’t need to be able to make multiple hops between Zigbee repeaters to reach the home WiFi gateway downstairs, if there’s already an additional WiFi AP on the top floor of a home.

To this end, the telcos could have a major role to play here, as they are the major provider of WiFi gateways to consumers. Adding a Zigbee radio to their designs would be a major coup for the protocol, and for the CSPs, could be quite low risk – as that 802.15.4 radio could handle Zigbee, Thread and Bluetooth.

In addition, Links points out the potential drawbacks of mesh networking, in terms of its power consumption when acting as repeaters, and the power budget used for network diagnostics. This would be considerably lessened if a device was only making one hop to the hub, which could then result in much longer battery life for these devices.

The discussion of mesh was covered in a recent Qorvo white paper, which you can find here. It’s an interesting topic to dive into, especially as one of the biggest problems facing the IoT in its infancy was the question of range and power consumption. Now, the network architectures themselves, in the home at least, have developed sufficiently to question some of the design assumptions and requirements behind the likes of Zigbee, Z-Wave, Thread, and Bluetooth Mesh.