By Michael C. Skurla
At the start of automation, which later was expanded with IoT devices to contribute to data, it was all about ‘now’. Real-time data allowing apps to warn people of problems or offer advice within the context of now. Though helpful, and certainly better than manually sifting through data, this was a very reactionary approach. People were still involved, though potentially less, and could be off-site to some extent. It was about monitoring and management.
We are now at a crossroads where analytics are playing more valuable roles, making data about ‘now’ still interesting, but what is actually more interesting is the historical context of the data from these data sources. This historical context, coupled with computing power brought to us from edge computing and data processing capabilities, enables a new context of data—that of the ability to learn and predict problems, and address solutions before they become the ‘here and now’. Hence AI and Machine Learning have brought predictive analysis to industry 4.0—including the value of IoT infrastructure, lowering the operational expense of managing infrastructure in a wide array of verticals.
The advent of the Internet of Things (IoT) redefined our relationship with data—as data became the new oil. In the first generation of IoT platforms it was all about the ‘now’. Alarming and event management helped people understand critical failures and problems without the requirements of boots on the ground—yet little was to be said about what to do with the historical context of the same data. Like refining oil, data refined allows you to extrapolate further actionable outcomes. Hence the second generation of IoT solutions redefined IoT’s role as no longer just supplementing an operation but becoming the analytical solution (and the digital boots on the ground) of the system.
The marriage of data infrastructure, edge computing, telecommunications advancements, and IoT technology has enabled a wide range of capabilities—from consumer applications to critical infrastructure upkeep. The analytics revolution is no longer confined to facility management. IoT ecosystems from edge, through fog to cloud, offer more than just the ability to monitor business processes and operations. What once was seen just as facility data, now can, and is having a context within a range of other activities for the business that are far outside of facility management. Marketing, logistics, staffing, safety; just to name a few. All from the same data, just analyzed with a different lens. Remembering that IoT ecosystems are only as good as the data model behind them.
IoT PLATFORMS NATURALLY FOSTER PREDICTIVE ANALYTICS
As part of advanced analytics, predicting future outcomes uses real-time and historical data together to look for patterns. What is key about an IoT platform is its ability to allow adaptation to these patterns based on operator contextual understanding. Explained another way—no single analytics platform knows all the trade secrets to operating a specific typical facility or portfolio of facilities. Yet an IoT platform allows for easy (nonprogrammatic) methods for an operator or business to put in known factors to look for about the operation, and then look for trends based on that input, to allow for predictive results. Typically, they perform on the fly configurable functions through the web interface allowing businesses to adapt to changes in expectation, understanding, or technology shifts.
Though IoT platforms offer a wealth of options for businesses to create their own predictive analytics, there is also a growing business of what has been coined as ‘microservice analytics’. These software companies still rely on the core IoT platform for data, but have special knowledge in specific verticals, and that of typical outcomes that would be advantageous—hence avoiding creating your own rules. Many businesses choose to do both internal and external microservice analytics. Either way, it’s all about saving money by automating people and process and allowing for a more cohesive remote operation which not only lowers OpEx (as first-gen IoT systems did), but ideally now adding value and revenue capability to the business beyond just ‘savings’.
EXPLORING THE IoT STACK
Effective IoT implementation depends on the choice of IoT stack (or platform) that integrates IoT communications and data flows but offers consistent security policies across the entire stack as well as comprehensive data storage and organization for proper data mining by analytics tooling (either internal or external). Since most organizations are challenged with IoT integration, IoT platforms ease the integration woes and challenges for IoT organizations by providing a full, end-to-end IoT solution–data acquisition, alarming and event management, internal analytics, often a scripting engine, reporting, and most commonly a robust cloud component to allow for organization and storage of data.
IoT-enabled platforms are the most cost-effective solutions to aggregate, organize, and harness insightful intelligence. With IoT devices and legacy systems all providing real-time data–typically a building already has a reasonably good amount of data being generated to work off for actionable analytics, reporting/alerting, asset management, production insights, etc. The IoT platforms offers the mechanism to harness all these systems (often from numerous vendors, generations of equipment, sometimes spread across large geographic regions) into one place to take advantage of this data effectively on scale.
PREDICTIVE ANALYTICS IS THE FUTURE OF IoT
Predictive analytics isn’t a recent phenomenon. It just has been difficult to automate. Tapping on aggregated data, algorithms and Machine Learning techniques, predictive analytics via IoT platforms now enable this without significant customization or proprietary integration and programing. The explosion of all forms of data generated by IoT devices and easy-to-use software now provides the ideal platform for organizations to tap into the gold mine of valuable analytics for actionable insights. Cybersecurity remains as a front and center concern for organizations, but pattern detection and behavioral analytics allows for network administrators to quickly detect abnormalities and advanced persistent threats that may lead to irreversible risks. Predictive analytics also allows organizations to boost operations and optimize resource management from forecasting inventory, determining equipment failures, preventive maintenance, mitigating risks/safety measures to securing quality and production failures. Overall, predictive analytics boosts improved business operations efficiency positioning enterprises with a competitive advantage that leads to increased bottom line.
A Forrester study shows that 90% of businesses expect data-driven insight to become a key differentiator by the year’s end–but need cloud-scale. The study shows how practical considerations around providing connectivity to remote locations, general suspicion about the security, capability, and trustworthiness of public cloud providers, have largely disappeared. With fewer businesses in the IoT space investing in building their own networks of data centers, the public cloud has become most ideal.
While initially companies involved in IoT projects were simply looking for monitoring capabilities, there’s now a demand for analytics, machine learning and AI. “Vendors must bake analytics, insight and action deeply into their platform offerings to support predictive maintenance, machine-learning powered workload optimization and scheduling, and more,” according to Forrester research. At a minimum, the analytics functions should handle analytics that include Descriptive—what’s happening real-time, Predictive–prevent unwanted and costly downtime and Prescriptive–provide information on ROI, new business models and ways to maximize output and efficiency.
A critical aspect of any IoT platform is its ability to manage the volumes of data generated and enable users with the ability to integrate actionable results, according to 451 Research. This also includes integration of data from across disjointed sites, systems, and infrastructure.
With an IoT data analytics platform that can mine, refine and process structured and unstructured data automatically, enterprises can make data-driven intelligent business decisions in real-time. Those platforms that offer a mixture of prebuilt tools allow end-users to customize their own business-specific analytics or support third-party analytics solutions. Gartner sees new delivery models, moving from system integration to insights-as-aservice, estimating that over half of data and analytics services will be performed by machines instead of humans by 2022. It also predicts that data, analytics and other technologies supporting them will increasingly live in edge computing environments, closer to assets in the physical world and outside IT’s purview. By 2023, over 50% of the primary responsibility of data and analytics leaders will comprise data created, managed, and analyzed in edge environments.
Michael Skurla is Chief Product Officer at Radix IoT, LLC. He can be reached at email@example.com.