In 2016, we saw breathless growth in the industrial IoT market. A bewildering mix of solutions confront the industrial operator, and many are choosing the easiest, most basic approach. In 2017, we see this customer base increasing in sophistication as they begin to reap early rewards—and stumble on pitfalls—from their early efforts. Here’s where the conversation is going in 2017 and beyond:

  • The cracks begin to show: Many vendors are chasing the $900 million industrial IoT market, but few have the necessary domain expertise to understand and incorporate industrial “things.” Most of these things—like trains, boilers, power generators—were not built from the ground up with security or Internet connectivity in mind. It takes significant experience to understand how to integrate operational technology—or OT—with IT and people to create end-to-end solutions for use in an industrial context. Soon we will see the vulnerabilities and gaps of early IoT solutions begin to emerge. Though plenty of customers exist that will buy basic IoT solutions, it will soon become evident that many players don’t have the domain expertise, investments, or patience to connect and manage these types of complex industrial environments. It’s easy to build software—it’s much harder to build a bullet train and then apply sensors and software to extract mission-critical insights that keep that train reliably running without downtime for maintenance or operational issues.
  • Predictive capabilities will proliferate: The industrial sector has been automating processes for decades using “machine-to-machine” or M2M technology. But things are changing—we’re moving from a reactive M2M causality approach of “if this then that” (i.e. if an RFID (radio frequency identification)-tagged crate of tomatoes passes my refrigerated truck reader, then it communicates the location and temperature to a logistics manager). Now we’re taking a more active, analytics-based approach that can help you stay ahead of potential issues, avoid unexpected costs/downtime, etc. The entire journey of those tomatoes, from farm to fork, can now be analyzed by multiple parties using open software that guides better decision-making by the farmer, warehouse operator, logistics provider, and retailer. And, when cross-referenced with adjacent data from soil sensors, weather forecasts, vehicle health, and check-out scanners, a predictive environment emerges.
  • People and machines are the same: Today, most IoT systems are designed to connect with machines; people simply consume information and use these systems via a portal. Down the road, I see our world evolving to an “Internet of Humans and Machines” supported by ‘digital twins.’ Digital twin is a term coined by NASA that models and represents a computerized companion to a physical asset, like the space shuttle in order to remotely monitor the health of its onboard systems. While most assets represented by digital twins are currently machines, there’s no reason why people shouldn’t be represented the same way by the same systems. With this model, people use wearables, glasses, or implants to connect to IoT systems on their behalf. Sometimes people convey aspects of their physical state directly using voice systems or by inputing information into a smartphone or tablet. For example, the digital twin of a patient and all her vital signs is represented visually on a device in a hospital without the patient actually having to be there. In this case, the constant flow of telemetry data ensures the digital twin is always up-to-date so doctors and nurses can monitor and diagnose their patient remotely as though she were there.
  • Analytics will inspire: I predict the industrial world will venture out beyond simple rules-processing and begin to care much more about advanced analytics, such as machine learning. Some say large-scale deployments of machine learning is still a few years off. This is due to a lack of machine learning tooling expertise and a dearth of data scientists. However, industrial customers will begin piloting these solutions in 2017. Machine learning is critical to the future of industrial operations, making it possible to avoid downtime by predicting equipment failures and saving money by stretching maintenance cycles through the determination of remaining useful life. This predictive technology will have a clear impact on a company’s bottomline that far exceeds the combined cost of sensors, connectivity, an IoT system, and analytics. For example, if a factory manager can’t predict the asset health of industrial robots on an assembly line, unplanned failures could shut the line down, leading to a loss of revenue plus a loss in confidence from from downstream distributors and customers.
  • IoT security moves to the forefront: While most organizations talk a good game about the importance of security, it often remains an afterthought in practice. Following the DDOS (distributed denial of service) attack in October 2016, it’s now top of mind like never before and industrial users will begin to take this more seriously in 2017. For more details see a post entitled, 12 Steps to Stop the Next IoT Attack in its Tracks. In addition to the safeguards outlined there, industrial users should put a special focus on how best to IoT-enable existing equipment that may not be securable as part of their security strategy. Implementing an IoT strategy doesn’t automatically mean your machines should break the bonds of your air-gapped factory to send telemetry and receive commands from over the Internet. Be pragmatic.

What do you anticipate for the industrial IoT in 2017?

Rob Tiffany is CTO, Lumada IoT platform at Hitachi Insight Group.