The great thing about standards is that there are always so many to choose from. We've seen the standards forest grow countless times before. The Internet of Things is a vast digital petri dish for them, and they just keep growing.
Recently we've acquired another initiative to pull together vendors working in IoT. It's called EdgeX Foundry, and it's the latest effort from the Linux Foundation.
Just don't call EdgeX Foundry a standards initiative, though, or you'll disappoint Philip DesAutels, the foundation's senior director of IoT. "We don't do standards," he says. "We do open-source software that defines effectively a de facto standard." OK, then.
The foundation's modus operandi is to create open-source reference software that others can draw on for their own implementation. It has an automotive Linux for car vendors to work from, for example, and its own blockchain implementation, called HyperLedger.
The Linux Foundation doesn't have just one de facto not-standard for IoT, though – it has several. DesAutels carves them up by target audience: industrial IoT and consumer. These two worlds approach IoT differently, he argues.
Industrial vendors each produce a tiny cog in a machine rather than a finished package. They want their single component to talk with lots of others so that systems integrators can work with them. Integrating a combination of obscure industrial controllers may only happen once or twice, on an ad hoc basis, making an open interoperability layer a useful way of cutting integration costs.
"EdgeX and the industrial world is looking at open frameworks," says DesAutels. "It's a common service layer that they can all point to so that their things work together and their target customers can assemble more complex systems out of components."
There are other efforts to harmonise connected components in the industrial space. AT&T, Cisco, General Electric, Intel and IBM founded the Industrial Internet Consortium (IIC).
"The Industrial Internet Consortium is absolutely everything about industry that there could be," says Ian Hughes, a former IBMer and now IoT analyst at 451 Research. This covers the intricacies of the production line, creating testbeds for these thousands of obscure components to work together. It will pull standards in and use them as necessary.
"EdgeX is about how you do this more complex processing closer as a component of your factory," Hughes adds. You have 200 robots. How do you build the code to control them all and where do you run it?
EdgeX would like to do it on a gateway device at the edge of the production line, but the software controlling those robots may also need to talk to the cloud in a distributed manner, he says. Yet another effort, the OpenFog Consortium, covers the distributed computing side of IoT.
A standard for consumer IoT
The Linux Foundation has experience on the consumer IoT side, too, although its journey has been more complex.
"The consumer side has been riddled with vendors trying to offer end-to-end solutions," says DesAutels. Vendors release their own frameworks and then muscle everyone else to join in. That's why Google released Works with Nest and why Apple has HomeKit.
Unlike the ad hoc integrations in industrial IT, in consumer IoT a single integration between a thermostat and a speaker with a digital assistant will be used in hundreds of thousands of homes. This has made an interoperability layer less imperative than dominance. If a vendor can grab a big enough share of those highly repeatable integrations by putting itself in the living room and in your pocket then others can come and work with it.
Consequently, the path to an interoperability layer in the consumer IoT world had been rocky. Qualcomm came out with the AllJoyn protocol at the 2011 Mobile World Congress. This was an interoperability layer that all consumer products could communicate with. Qualcomm gave it to the Linux Foundation in 2013, which promptly used it as the basis for its AllSeen Alliance.
At around the same time, Intel launched the Open Interconnect Consortium, which competed with Qualcomm, in an attempt to wrest control from the AllSeen Alliance after some companies in the OIC voiced concerns about Qualcomm's role in it. The two parties eventually resolved their differences, and in 2016 the AllSeen Alliance and OIC merged to create the Open Connectivity Foundation.
This leaves the Linux Foundation still managing AllJoyn, which continues as a joint project with the OCF, along with IoTivity, which was the OIC's equivalent interoperability protocol.
"They've been working hard since the merger to bring those projects together, and we’ve been seeing results in terms of software bridges and interoperability between IoTivity and AllJoyn," he says. "That's now one community."
It's a community with two parts, though, because the OCF manages the standardisation of these things, while the Linux Foundation manages the software projects.
Standards, standards, everywhere
Alongside these initiatives, there are also others operating at various levels. The IEEE P2413 project is working on a standard for an architectural framework for the IoT, which will bring together what it sees as fragmented efforts in various verticals. It will draw on existing standards and projects, and there will be a reference architecture.
While some IoT initiatives focus on application-layer connectivity between IoT devices, others come from the networking end of the conversation. In summer 2016, chip vendor Sigma Designs, the main backer of the Z-Wave wireless mesh-networking stack, opened its Z-Wave interoperability layer up for public access. Previously only Z-Wave Alliance members could get it. This included the API specification for Z-Wave over IP, and its Z-Ware middleware product for gateways, along with the Z-Wave S2 security application framework.
The ZigBee Alliance, which manages the rival Zigbee standard, announced the Dotdot interoperability layer at CES this year, calling it a "universal language for IoT".
Both these efforts compete with an existing initiative, called Thread, backed by Google's Nest, along with Yale Security, Silicon Labs, Samsung Electronics, Freescale Semiconductor, Big Ass Fans and ARM. That entered the ring in 2014 and now also counts Qualcomm – the originator of AllJoyn – among its backing brands. Further down the networking stack, we have specifications for low-power wide area networks that will enable electron-starved IoT devices to work together remotely. The 3GPP has created Category M1 and Category NB1 for this purpose, while LoRa is another.
Why is this space so cluttered? It's about vendor politics – and it always has been.
"As a rule, the larger vendors will jostle for positioning in the standards efforts. Startups will try to carve out a niche but will eventually fall in line with whatever Cisco, HPE, Google, Apple, Amazon, Samsung, etc. hedge their bets on," says Chris Wilder, senior IoT analyst at Moor Insights and Strategy. "These larger vendors will pursue and maybe double down on several standards in order to eventually pick the winners and losers."
DesAutels admits that the space is fragmented. So which way should vendors jump? "I don't know that if you're a vendor coming into the space there is one answer right now," he says, admitting that the AllSeen Alliance/OCF effort isn't delivering anything notable yet. "It would be a complex decision space. What radio are you using? Which community do you want to be a part of?"
And in the darkness bind them
Will we ever arrive at one IoT standard to rule them all? Wilder has identified at least 90 of them, but believes everything will one day just run over IP, with application integration services like Zapier and IFTTT disseminating real-time and multi-dimensional data that people can do useful things with.
"The overall value and revenue opportunities for IoT are through value added services, not standards or ecosystems," he says.
The lower levels of the IoT networking stack may become less important if Hughes' predictions come true. He draws allusions with the journey towards 5G in the telecoms world, arguing that these companies really just want to connect in whatever way is most appropriate at the time based on the type of data being sent. Streaming VR data would need one kind of connection, while telemetry data from a car would need another.
"It becomes this sliced network," he says. "You just say 'this is the information I need from these devices and this is the quality of service, so let the intelligent network sort it out'. We're not there yet."
The discussion will evolve from transport mechanics through to loftier goals. "They are going to evolve into more complex orchestration platforms that decide for you which network slice it will use, and how it is going to arrange that – where the processing is going to occur," he says.
That's an ideal future, but right now we're stuck with different interoperability options at all layers of the stack. Users may gravitate to a smaller number of interoperability protocols over time, but unravelling the standards spaghetti will be too difficult, warns Andy Castonguay, principal analyst at Machina Research. Aside from all the newer equipment there's the issue of connecting legacy equipment to consider.
"Many IoT platforms are so agile that developers and integrators can add 'connectors' for most additional protocols and standards within a day or so," he says.
What's most worrying is that a large proportion of the existing IT kit out there may itself become legacy as standards discussions evolve. Let's just hope that your smart thermostat is smart enough to accommodate some fairly hefty firmware updates. In the future, it may need them. ®