It’s been nearly 15 years since my first Internet of Things (IoT) project, which was a co-development effort with Rockwell Automation focused on industrial ethernet switches. As we evolved our work to connect more and more industrial applications, machines and processes, we became part of the birth of the IoT and, more specifically, the Industrial Internet of Things (IIoT).
From the beginning, the big benefit of IIoT has been to bring together industrial, IT and operational technology processes into a single system and automate where possible to improve operational efficiencies and cut costs. Production efficiency and agility were certainly what iconic motorcycle manufacturer Harley-Davidson was looking for when it became an early adopter of industrial IoT.
The company had many challenges: IT wasn’t aligned with production; labor was too expensive; and data was incompatible and unusable. So, it streamlined separate systems onto a single IP network, consolidated data and fully enabled one factory with IoT capabilities. As a result, decision-making speed was improved by 80 percent, build-to-order cycle time was reduced from 18 months to two weeks, and production throughput increased by nearly 7 percent.
We can see similar successes in every industry: Ford Motor Company has enabled 25 of its 40 assembly plants with IoT technology to speed communications, improve scheduling and manage more than 2 million production variations in real time. Cisco reduced energy consumption by 15 to 20 percent in one manufacturing facility by installing sensors to track energy flow and identify underperforming equipment. PepsiCo harnessed a unified IoT platform to improve reliability of its manufacturing system, reduce downtime and streamline communication through remote monitoring.
It’s clear that these kinds of improvements in productivity and cost savings are a major driver of IoT adoption. But companies and industries that stop there are leaving money on the table.
Edge Computing Index: From Edge to Enterprise
According to a study by Futurum Research, nearly 64 percent of respondents are focused on combining edge computing and data center analytics; more than 15 percent aim to keep edge computing and data center analytics separate; and more than 20 percent are not sure whether to combine them or keep them separate.
The untapped value of IoT data
The real value of IoT lies in its data. In fact, the main reason IoT devices are connected is to generate big data that can be analyzed and acted upon when needed by other devices, applications, machines or people.
Connected industrial systems can generate massive amounts of data from cameras and sensors measuring temperature, pressure, moisture, speed and virtually every other aspect of operation. However, more than 99 percent of this data goes unanalyzed and unused. The data that is analyzed is usually collected and sent to the cloud for batch processing and report generation. That’s fine if you want to analyze trends over 30 years of seismic data. But if you want to take advantage of applications that use streaming data and real-time data analytics to make split-second decisions, you must turn to an end-to-end architectural approach from the edge of physical operations all the way to the cloud.
For example, what if a temperature sensor on an oil rig shows that a drill bit is heating up? You want to be able to pull that piece of information out of the data stream, identify it as an exception, analyze it to see what’s going on with the drill, and recommend actions to take to prevent a disruption to operations. And you need to do it in near-real time, before a part breaks and damages other equipment. Better yet, what if you could analyze all your data in the context of historical data from similar equipment across the industry and be able to predict when a drill is likely to fail? This is where IIoT data becomes worth its weight in gold.
The number of sensors generating IoT data is growing at an astonishing rate – analysts have estimated that by 2020, 40 percent of all data will come from sensors.1 A single offshore oil rig generates up to 2 terabytes of data each day. A jet engine might produce a terabyte per flight. The challenge is to identify and process the relevant data points hidden in vast streams of routine or unimportant data from IoT devices – and to do it in near-real time to generate immediate insights, recommendations and actions.
To unlock the full value of IoT data, many organizations are moving intelligence and processing closer to the device itself, where the data originates. Sometimes these capabilities are pushed directly into IoT devices or aggregation points (i.e., IoT gateway) at the “edge” of a network (i.e., edge computing). Or, fog computing pushes processing and intelligence to a local area network. Both approaches solve issues with latency, bandwidth, reliability and security that have traditionally limited the performance and functionality of IIoT solutions.
Consider the oil rig example. By extending existing cloud architectures to an oil rig, organizations can use fog and edge computing to enable real-time data to be processed and analyzed locally based on policies coming from the cloud. Then, only exceptions and alerts would be sent over the satellite link.
Edge and fog computing support transformational IoT applications that depend on being able to access and analyze data in real time. Autonomous drones, for example, need real-time data analytics to choose the most efficient flight path and react instantaneously to avoid bad weather, trees or power lines. Combined with artificial intelligence capabilities, a drone can even operate in dark, obstacle-filled environments beyond the reach of the internet and GPS. Such drones are ready for high-value, mission-critical applications, whether that means an inspection of a gas pipeline or secure package delivery in New York City.
AI and machine learning make data smarter and more valuable
IoT and AI have a remarkably synergistic relationship. You might say that AI is the brain of the business and IoT is the body. AI, especially machine learning, provides intelligence – the ability to evaluate options, learn from experience and make smart business decisions. IoT, like the body, provides the ability to sense and act. IoT delivers both the data AI needs, and the physical means to act on AI’s decisions.
The convergence of AI and IoT is creating countless new opportunities and business outcomes. In manufacturing, predictive analytics gives production managers the intelligence to evaluate the trade-offs between building a new plant, for example, or buying extra capacity as needed. And preventive maintenance systems use IIoT data plus AI to predict and prevent equipment problems before they happen, saving millions of dollars in unplanned downtime costs. For example, Western Digital uses asset performance analytics to identify potential failures early in the production process and make timely decisions to avoid yield excursions. As a result, the company has lowered the overall number of returned units, boosting customer loyalty and trust – which has a direct bearing on its revenue.
A tipping point for IoT value
As AI, edge and fog technologies mature, they are accelerating the Internet of Things toward a tipping point – moving from creating incremental value to creating entirely new value propositions, business models and even industries. There are already abundant examples of new value creation at the intersection of these technologies.
Consider new value propositions, such as mass customization and personalization. With IoT and automation, customers can order a car, a suit or just about anything else, specifying desired options – and it rolls off the production line made to order at a cost close to that of mass-produced goods. For example, Daihatsu Motor Company is using 3D printers to offer customers 10 colors and 15 base patterns to create their own “effect skins” for car exteriors.
IoT has emerged as a foundational capability that (when combined with machine learning and fog or edge computing) is creating brand new industries such as autonomous drones. It is also a key force behind the convergence of existing industries such as transportation and technology, and retail and manufacturing.
The collaborative, connected nature of IoT is also ushering in new business models, including the “co-economy.” This model is based on dynamic ecosystems of partners and customers that bring together complementary strengths to deliver co-created solutions. For example, GE Transportation has integrated SAS® Analytics capabilities with its cloud-based operating system for the IIoT to give customers analytical insights on their data in real time. The solution deciphers locomotive IoT data and uncovers use patterns to keep trains on track.
These and many other examples are brilliant glimmers of what is right around the corner. IIoT data is a rich mother lode of value waiting to be mined. Together, technologies like edge and fog computing, machine learning and AI can unlock its hidden value.
SAS and Cisco: Transforming information into insight
More data is being generated than ever before. Do you know what your data is telling you? With SAS and Cisco, you can manage and analyze large volumes of data in minutes or seconds, not days or hours.
About the Author
Maciej Kranz brings 30 years of networking industry experience to his position as Vice President of Cisco’s Strategic Innovation Group. In this role, he leads efforts to incubate new businesses and accelerate co-innovation internally and externally with customers and startups through a global network of Innovation Centers. He pioneered dozens of IoT projects across multiple industries, publishes an IoT newsletter, spearheads an industry leadership community and wrote the New York Times best-seller Building the Internet of Things.
- AI in manufacturing: New oppportunities for IT and operationsAn AI survey reveals that leaders and early adopters in AI are making important advances and are identifying and expanding on what works as they use AI in more ways and more parts of their organizations.
- A guide to machine learning algorithms and their applicationsDo you know the difference between supervised and unsupervised learning? How about the difference between decision trees and forests? Or when to use a support vector algorithm? Get all the answers here.
- Meet the data scientist: Kristin CarneyWhen Kristin Carney graduated with a BS in mathematics, she wasn't sure what she wanted to do with her degree. That’s when she began researching data science.