Once 46 billion devices, sensors and actuators are connected by 2021 – according to Juniper – we'll be well on the way to a bigger, more powerful and much more settled Internet of Things (IoT). So what will an IoT 2.0 look like?

Interoperability and open ecosystems

One of the key characteristics of the IoT 2.0 will be common standards. The IoT umbrella is vast, and the many industries it covers – from factories and automotive through to building automation and networking – each have their own protocols, interfaces and hardware.

“The real paradigm shift in IoT has been the availability of development boards at a fraction of the cost of old proprietary solutions,” says Thibaut Rouffineau, Head of Devices Marketing, IoT, Phone, PC at Canonical. He thinks that in the light of falling hardware and development costs, many companies are re-evaluating their tech choices. “They’re moving away from proprietary non-interoperable solutions and towards more open ecosystems and software.”

Happily, we’re now on the cusp of common industry standards.

IoT-enabled commerce will embrace machine learning

Contextual commerce

For retailers and e-commerce, the IoT is already massive – and growing fast.

“IoT-enabled commerce already exists, but we can expect to see an increase in connected devices using machine learning to predict what users want by analysing historical and current data,” says Jim Bailey, senior managing director at Accenture Mobility.

He thinks that autonomous ordering is also a possibility. “If rules have been set, they might even autonomously order and pay for replacement parts or replenishment of groceries, without the need for any human intervention whatsoever,” he says.

Cognitive-first apps

The IoT will have to change drastically just to cope with the sheer number of devices around. “IoT 2.0 will be defined by the immense number of IoT devices, sensors and actuators, deployed in very diverse environments,” says Dmitri Tcherevik, Chief Technology Officer at Progress. He think that IoT platform architecture will need to evolve from the current hub-and-spoke model towards a more distributed peer-to-peer model.

“The devices will also have to be given greater autonomy, which will inevitably lead to greater cognitive, adaptive, and predictive capabilities both at the individual device level and at the IoT platform level,” he says.

The result is likely to be machine learning algorithms, more fog/edge computing, and new APIs. “This will give rise to a new class of cognitive applications that will put the power of IoT in the hands of the end users,” says Tcherevik. These cognitive-first apps will teach machines to automate data and improve the accuracy of failure predictions for IoT-connected equipment.

Are we close to getting £1 supercomputers-on-a-chip into IoT devices?

Explosion in low-cost IoT devices

Hardware is getting cheaper, and compute is getting more powerful. “At the point where anyone can put a £1/$1 supercomputer-on-a-chip into their IoT device, which may only be a few years away, development of custom hardware will seem like a moot point,” says Rouffineau. At that point, why should IoT device manufacturers develop their own semiconductors when all the features they need can be software-defined?

Could Bluetooth radio chips become ‘smart dust’ painted on planes? (Image Credit: Cambridge Consultants)

Others are working on reducing the size and cost of Bluetooth chips so drastically that connectivity, too, becomes software-defined. Currently the most expensive part of Bluetooth Smart chips at the heart of every connected IoT device is the radio, but Cambridge Consultants’ Pizzicato all-digital radio tech slashes the size by 90%.

“We will kick-start the IoT by being the first in the world to create a fully digital radio transmitter,” claims Rob Milner, head of smart systems at Cambridge Consultants, predicting that Moore’s Law means that sensor-packed ‘smart dust’ could even be produced as paint on airplanes, or sprayed onto mountains to act as an early-warning system for avalanches. That's pure IoT 2.0.

  • Here’s why the Internet of Things could fail

Greater focus on security

As IoT hardware becomes cheaper and more powerful, security in the next phase of the IoT will be critical. “Even relatively simple, everyday IoT devices have proven themselves capable of launching DDoS attacks on scales hitherto unimagined with simple GRE floods, as we saw in the case of the Mirai botnet attack,” says Rouffineau. “Imagine if those devices were more sophisticated, more powerful and more ubiquitous.”

He thinks that IoT security will need to be addressed at the device OS level. “The industry needs to adopt a more standardised approach to the IoT OS it uses, centralised and digitally-signed software updates, and an assumption that apps and devices are ‘guilty until proven innocent’,” Rouffineau asserts.

Authentication and identity

Static user IDs, pass-phrases and passwords are dead; IoT 2.0 wants nothing to do with them. Cue contextual and continual authentication that uses data such as geolocation, device fingerprinting, and time of day login patterns to authorise access. But it goes much further than this.

“Most IoT devices will require a non-interactive method to login, so the use of cryptographic materials such as certificates and private keys becomes ever more necessary,” says Simon Moffatt, Senior Product Manager at ForgeRock. “Underlying APIs and cloud services that IoT devices will access, should be protected via throttling and rate-limiting gateways that are continually checking the status of previously issued sessions to make sure they are not being used maliciously or have been stolen.”

Security will surely be at the heart of any IoT 2.0.

5G will need software-defined networking if it’s to help the IoT

5G and software-defined networking

Able to support a massive amount of devices per square kilometre and capable of ultra-reliable and low-latency communications (uRLLC), 5G networks are heralded as a step-change for the IoT. The reality is very different.

“5G alone can’t support the IoT – 4G today is unable to do so due to a lack of blanket coverage and inability to support the demands of data-intensive applications such as HD and 4K video,” says Sean Collins, Regional Director for UKI and BeNeLux at Extreme Networks. “For IoT to evolve, a highly managed mixture of technologies including 5G, wired and wireless using SDN (software-defined networking) will be the best way to prioritise, deliver and manage IoT applications.”

IoT 2.0 will be a jigsaw of all kinds of connectivity, and both old and new devices – and that makes retrofitting a key component.

Retrofitting legacy devices

An emerging way of retrofitting legacy devices is using data-over audio; small, inaudible bursts of ultrasound that devices can use to exchange data.

“Instead of using radio frequencies we encode information in a series of audible tones that can be played by any device’s speaker and received by any device’s microphone,” says Dr Daniel Jones, Chief Science Officer at Chirp, whose technology generates dynamic data on-the-fly to produce unique tones, and requires only a mic and a speaker to work.

Chirp’s tech is being used by some companies to retrofit legacy devices to connect them to the IoT, and help devices communicate with each other without requiring internet access (it has been trialled at nuclear power stations in the UK, where internet access is restricted).

“Sound is almost the lowest common denominator because it helps cheap archaic devices communicate,” says Jones. “So it could be a key part of the jigsaw in environments where interoperability is key.”

Chirp retrofits legacy devices using data-over audio

When will we see IoT 2.0?

“Before we can talk about an IoT 2.0, we need an IoT 1.0 characterised by large deployments and successful business models, but we’re not there yet,” says Rouffineau. The IoT needs to first settle, define its standards, improve its security to protect privacy, and iron out business models. “It’s also likely that software definition, and the use of IoT ‘apps’, will play a big part in the creation of a first successful IoT era,” adds Rouffineau.

The pulling together of disparate data is what’s characterising current IoT 1.0 projects, but there’s something missing. “They lack that ‘helicopter view’ where they can understand all that data in a single pane of glass that will allow them to really make gains on operational efficiency and cost-effectiveness,” says Collins. “That’s why it’s so important that we look where we’re going when building infrastructure, making sure we are able to support the bandwidth and usability demands five years from now, as well as today.” 

It might be five years distant, but IoT 2.0 is on its way, with device miniaturisation, better power efficiency and connectivity, more sophisticated system architectures, and new machine learning algorithms all in the pipeline. Says Tcherevik: “IoT 2.0 is all but inevitable.”

  • We’ve highlighted 10 surprising trends in the IoT for 2017