What Is Edge Computing?
Location, location, location: it’s not just real estate anymore.
During the cloud boom over the past decade, we’ve seen organizations across the board migrate much of their on-premise infrastructure to remote data centers. While this new architecture has proved pivotal for applications spanning from collaboration tools to big data analytics, we’re now seeing a shift in the other direction: many companies, especially manufacturers, are bringing their computing power back into their in-house devices.
This time, however, it’s not servers or desktop PCs. It’s the industrial internet of things (IIoT), a collection of interconnected smart devices, equipped with sensors, actuators, and an embedded hardware/software stack that facilitates everything from production line automation control to machine-to-machine (M2M) communication protocols. This class of devices includes robotic arms, flow control systems, environmental controllers, and more.
These devices rely on real-time operability, fault-tolerance, and an increasingly heavy compute workload to function. That’s why, instead of sending data to some off-site server or even a centralized control hub, industrial control engineers are employing recent advances in edge computing technology to enable these IIoT devices to ‘think for themselves.’
Edge computing means bringing processing right to the data source. And industry is taking note. As Businesswire reports, “The global edge computing market is projected to grow from USD 3.6 billion in 2020 to USD 15.7 billion by 2025, at a compound annual growth rate (CAGR) of 34.1% during the forecast period.”
Edge Computing Advantages and Use Cases
Photo by patricio davalos on Unsplash
We’re seeing this shift towards edge computing for a few key reasons. First, as Christian Johansson writes for Control Engineering, “Putting the computing power near the devices it serves produces one major obvious benefit — enhanced speed by reducing latency, the time needed for data to travel from source to destination. Latency can be greatly reduced by edge computing compared to sending it to the cloud.”
For operations that demand nothing less than real-time control, reducing latency is critical. Even delays of a few milliseconds can throw off a highly calibrated assembly line. Second, edge computing saves bandwidth. While this may not be a huge concern for plant floors where IIoT devices can connect to a LAN via Wi-Fi or Ethernet, it’s a major factor for devices that rely on longer-range connectivity options like cellular or satellite, such as those found on off-shore rigs or other remote locations. At a certain point, sending constant streams of data to and from the cloud just isn’t feasible.
We’re already seeing companies utilize edge computing to reap these advantages for specific use cases. For instance, additive manufacturing for complex, custom parts like airplane engines benefits from local compute. “With edge technology, data is collected and analyzed at the point of operation to determine whether the printer is aligned properly prior to production,” writes Beth Stackpole for Automation World. This allows “any necessary adjustments to be made in real time so costly rework is avoided.”
Another use case that Automation World’s Mario Ishikawa points towards is integrating machine learning (ML) into supervisory control and data acquisition (SCADA) systems. These systems require low latency and high availability, making them unsuitable for cloud computing. However, by using microcontrollers to process data and edge GPUs to support the AI, manufacturers can create SCADA systems that are better at detecting errors and points of failure, especially before they become a problem.
Artificial intelligence and 3D printing are just a few of the innovations pushing manufacturers towards increasingly computing at the edge. 5G for IIoT, M2M communication like MQTT, and, of course, autonomous robotics are some other examples of technologies that rely on robust edge computing.
Conclusion
Ultimately, these compute locations live on a continuum, and, depending on what we want to do with our data, we’ll pick the right spot to run our workloads. As a general rule, the more immediate the application, the closer to the edge we want our computing to be.
One thing’s for certain. As manufacturers push towards the smart factory model of industry 4.0, edge computing will remain a major role-player for turning this vision into a reality.
Published By Zachary Stank
Originally published at https://www.linkedin.com.
