What is Edge Computing?

what is edge computing large

If you are a line engineer, or anyone involved in improving throughput in a manufacturing or logistics operation, edge computing is definitely a technology you should pay attention to. It has the potential to give you better insight into the operations of your line as well as the ability to make faster and more accurate decisions.

As the peripheral devices and sensors that get lumped under the overall term Internet of Things (IoT) generate more data and demand more network and cloud resources, two problems emerge:

  • Bandwidth is limited, and transferring large amounts of data becomes expensive
  • Round trip time delay, or latency, leads to slowed decision making for time-sensitive operations

At the same time, devices on the factory floor, including highly capable smart cameras and barcode readers, are acquiring more and more onboard computing power, and so, are generating more and more data. By moving data processing closer to the data-gathering devices themselves,—that is, to “the edge”—you can eliminate the latency of sending data up to the cloud and waiting for instructions back, reduce network congestion, and increase reliability.

Edge computing as part of the production information system

Typically, data generated by sensors transmit via a gateway to centralized cloud applications such as a manufacturing execution system (MES), enterprise resources planning (ERP), and a wide range of other line-of-business and operational software. This information enables factories to generate and implement valuable efficiency enhancements. Distributing all that data, modeling it, and then running analytics requires a lot of computing power in a centralized system.

With edge computing, the smart devices on the line can filter and compress their data, decreasing the load on the network. At the same time, they can provide controls engineers with real-time visibility into the operations of the production line, giving them the ability to have a real effect on improving overall equipment efficiency (OEE).

Challenges and opportunities on modern production lines

Keeping today’s complex production and distribution systems running efficiently and reliably is a difficult task, and machine vision systems have been helping automate these key manufacturing processes for years. Success of these facilities is often measured on throughput, and any disruption to these process flows carries heavy costs and penalties. As a result, quick action is necessary to resolve issues and minimize downtime, as minutes of lost production or shipments can cost thousands or even millions of dollars.
However, production managers often lack the right system performance data to adequately diagnose issues as they occur and are therefore left guessing when issues arise. The performance monitoring information that they do have is often restricted to overall averages instead of showing specific or trending issues. 

lack of visibility into system performance

Lack of visibility into line operation leads to:

  • Unplanned downtime
  • Deferred or unnecessary maintenance
  • Inability to identify and pinpoint where waste occurs
  • Failure to identify causes of rate drops and errors
  • Inability to measure, benchmark, and improve performance over time
  • Failure to increase throughput
  • Ineffective communication to plant management

In addition, device management can be cumbersome as operations scale up. Without a system to track even minor changes to settings, negative impacts to overall performance become harder to diagnose. Edge computing puts more information and analytics in the hands of line engineers and extends their ability to understand and control the lines they manage.

Edge computing in action: identifying unauthorized changes to readers

An example of a specific edge computing application can help clarify the benefits. A key advantage of edge computing is the ability to easily drill down to understand if the issue is at the line or reader level to see if there is some kind of outlier in code quality, monitor contrast, code position, or other metric. This kind of root-cause identification can expose some too-common misbehavior.

In a busy plant running multiple shifts with multiple lines, and with reporting requirements for excessive no-reads or delays, there can be a strong temptation to tweak an offending scanner to lower that no-read rate below the threshold, on that line, for that shift. The next day the first shift engineer will see that something has changed but be unable to figure out exactly what change was made and when. A quickie solution for someone’s specific problem has led to a more significant slowdown overall.

With edge computing, the engineer can go in, see the changes, when they were made, and the effects of the changes. They can revert to the previous state. After that situation is fixed, a more extensive analysis can then be run to see what might have been causing that high no-read rate. Addressing the root cause will improve OEE and also minimize the temptation for ad hoc camera adjustments in the future.

More power to the edge

Edge computing provides enterprise applications with the cleaned and structured data to make effective long-term and plantwide decisions. It also provides engineers on the line with the ability to optimize and improve operations on a day-to-day basis. For more information, download the Edge Computing Whitepaper.

More Posts on

GET ACCESS TO SUPPORT & TRAINING FOR PRODUCTS & MORE

Join MyCognex