Intelligence at the Edge: Smart Sensors Will Need to get Much Smarter

By Dick Slansky

Intelligent sensors are one of the essential components of IIoT. What this also means is that more intelligence will be required at the edge for the physical devices, machines, and equipment that manufacturing OEMs need to run production systems and to provide service in the field. These physical devices and sensors are indeed the “things” in the IIoT that will provide the data that drives the function of the connected ecosystems and are at the edge of the digital thread.

Today, manufacturing OEMs regard production data and maintenance and operational data as one of the major sources of big data in both the manufacturing environment and service in the field. Further, most of this data is unstructured, meaning that it must be converted into meaningful and actionable information before it can be applied to areas like condition monitoring, predictive analytics, and operational intelligence applications. As we move up the IIoT stack from the edge devices to connectivity, to edge computing, the data is transformed through data element analytics, aggregation, and levels of normalization. In order to implement concepts such as the digital twin, and to apply predictive and prescriptive analytics, it will become necessary to process more of the unstructured data at the edge device rather than just accumulate data at edge computing levels. That is, more unstructured data will need to be transformed into actionable information at the edge device/smarter sensor interface so that higher level analytics and operational intelligence platforms can work with information rather than unstructured data. What this means, quite simply, is that smart sensors are going to become an order of magnitude smarter than they are now.

The IIoT and the ecosystems that will emerge, will provide platforms for devices to generate and share far greater quantities of data than ever before in industrial environments, and enabling more sophisticated control and management of processes, machinery, and maintenance schedules. Traditional data gathering approaches such as SCADA, in which passive sensors channel raw data back to a central controller are already giving way to IIoT solutions that can offer faster response time, more efficient data gathering capacity, and big data services such as predictive maintenance and autonomous self-optimization.

Analytics performed in the Cloud are able to identify trends and patterns that human operators or industrial analysts unable to detect, generating higher level intelligence that enables processes, equipment, factories, and the digital enterprise to operate more efficiently and cost effectively. Interestingly, even with a significant amount of advanced analytics taking place in the Cloud, the essence and effectivity of the digital twin concept will be measured by the degree of intelligence at the edge.

Reprinted with permission,original blog was posted here

About ARC Advisory Group ( Founded in 1986, ARC Advisory Group is a Boston based leading technology research and advisory firm for industry and infrastructure. For further information or to provide feedback on this article, please contact

Share This Post

Leave a Reply