Whether you call it the Industrial Internet of Things (IIoT), Industry 4.0 or machine-to-machine (M2M) connectivity, there’s a trend that’s impossible to ignore: the manufacturing sector is getting connected, and fast. With slow economic growth and a volatile global market, competition is increasing and a thirst among industry for assets connected by wireless or wired networks will create an IIoT that will dwarf the consumer IoT.
However, to realise the predictions that sensors deployed in IIoT will comprise 64% of total sensors used in the Internet of Things as a whole, upgrades are needed. And we’re going to list 10 of the most important ones in this article.
Forget smart fridges – here’s why the Industrial Internet of Things is the real revolution
Since every device needs a globally unique IP address to get online, for the cloud to grow big enough and fast enough to cope with a burgeoning IIoT, Internet Protocol Version 6 (IPv6) will need to catch on. Whereas IPv4 permits four billion IP addresses, IPv6 creates a much, much larger address space of many billions, therefore allowing exponential growth.
“IPv6 is very important as it will allow the IoT to scale to accommodate the billions of devices everyone is predicting will need to be connected to the internet,” says Lis Strenger, IoT and Emerging Technologies at Red Hat.
With millions of IIoT devices predicted to arrive in the next few years, the public cloud will need to be able to store, manage and analyse a lot more data. However, which parts of the IIoT will see the most impact from rapid growth remains to be seen.
“It’s still too early to see all of the patterns that are emerging around Industrial IoT use cases, and how they stress cloud infrastructure,” says Strenger, “but it’s clear that the infrastructure for handling data will have to change to handle size, various data formats and streams, inconsistent frequencies, and requirements around security and compliance.”
For some, the biggest challenge facing a growing IIoT is hackers and malware. “We simply cannot afford to architect the next generation of IIoT systems without serious considerations of security,” says Dr Hamed Soroush, Senior Research Security Engineer at Real-Time Innovations, and Industrial Internet of Things Consortium (IIC) Security Working Group Co-chair. “Risks range from huge financial losses, such as having parts of the power grid down for even a few hours, to endangering human lives from attacks on medical devices, or on industrial control systems.”
Such security breaches in the energy and manufacturing sectors have already happened, including cyber-attacks on three Ukrainian energy companies in 2015, and on a New York dam by Iranian hackers in 2013. “We also need to protect the existing legacy systems from security attacks and incidents, and that is a challenging task,” adds Soroush.
Getting online to send sensor data might be the ultimate goal of devices in any IIoT system, but short-range technologies will also have a huge role to play. Not by sending data over long distances, but by daisy-chaining devices over vast areas, creating mesh networks that link any connected assets to each other.
If there is to be a real-time IIoT masterminded by the cloud, the development of 5G cellular connectivity will almost certainly be a requirement. “IIoT solutions that require real-time currently do not typically connect to the cloud because of the latency introduced by a round-trip between device and cloud,” says Strenger.
Although the kind of latency offered by 4G (which transports data in 55 milliseconds) is fine when collecting sensor readings for batch analysis, for automation in critical situations where real-time reactions to data are needed, only 5G’s sub-1ms latency will do.
IIoT systems may collect a lot of data, but it’s fragmented and needs distilling. “One of the reasons to implement an IoT solution is to collect data to analyse for additional insights into operations [for] improved production quality, predictive maintenance of equipment, and better utilisation of resources,” says Strenger. “Many companies are finding new uses for the data they collect, for example offering their customers new information resources.”
With the appearance of 5G networks, real-time big data analytics will become a possibility.
The IIoT has a lot of network protocols all vying for dominance. From Wi-Fi, Bluetooth 5, ZigBee and all kinds of cellular standards to Z-Wave, 6LowPAN, Sigfox and LoRaWAN, the sheer number of network protocols available mean devices are often isolated and incompatible with other IT infrastructures.
There are specific scenarios for each (for example, a low-power device that only sends a status update once per day requires only a low bandwidth Low Power Wide Area Network), but expect some protocols to become dominant, and others to disappear as developers begin to think in terms of systems, platforms and their interconnectedness, and not just how to most efficiently operate a single standalone IIoT project.
The IIoT might be a shiny new concept, but legacy tech is everywhere – and often acts as a brake on designing fresh projects. Consequently, IIoT systems are typically the convergence of IT and OT (operational technology). “Such convergence calls for two requirements,” says Soroush, who underscores the need for new architectures, protocols, software, hardware, and processes for supporting greenfield deployments, but also for the integration of legacy systems.
He continues: “Most of the OT technology used today is showing its age, but they are often hard to replace or upgrade due to a variety of reasons including high availability and high reliability requirements, high costs, and the need for recertification in highly regulated sectors.”
With the average lifespan of OT devices being around 19 years, this is a problem that’s not going away anytime soon.
“One of the big drivers for the IIoT is advanced data analytics that target data-driven optimisation of industrial processes,” says Soroush. “Machine learning plays a massive role in that respect, and it will also play a significant role in introducing further automation to industrial systems and processes.”
Machine learning – the science of giving computers an ability to learn from data and take actions without being specifically programmed to do so – can also include computer vision and object recognition. That could inform robotics and other kinds of automated processes, such as machines that can identify faults on a production line, or specific components needed for assembly.
Although a big reason for the IIoT’s development is to automate factories and production facilities, the increasing complexity of the IT backbone will itself require an automated platform.
“Automated platforms will be important to the IIoT as soon as companies have real production scenarios with thousands or millions of devices and gateways to manage,” says Strenger. “At that time, they will need a management strategy that automatically takes care of software updates and upgrades, patches, deploying new features, and functionality.”