5G and the factory of the future

The old adage “build a better mousetrap and the world will find its way to your doorstep” is often cited as a metaphor for the transformative power of innovation.

Today, that saying might as well be “build a better manufacturing plant and the data will find its way to your doorstep.” The growing adoption of private 5G wireless networks, along with the accepted concept of edge computing, is enabling manufacturers to gain real-time, actionable insights that are transforming the industry in safety, worker safety and security. quality.

In a study conducted just before the pandemic by Deloitte and the Manufacturer’s Alliance for Productivity and Innovation (MAPI), respondents cited double-digit gains in labor productivity (12%), factory capacity utilization ( 11%) and total production (10%) for the previous three years thanks to their smart factory initiatives. With 5G, these pioneering manufacturers will have the ability to connect more devices and collect more new types of data, keeping them at the forefront of the industry.

When combined, 5G and edge computing provide the basic infrastructure for new ways of working. Although Wi-Fi is strong enough to support multiple devices in a home or office, it has never been well suited for the hundreds or thousands of sensors and other devices in a manufacturing or warehouse environment. Likewise, hard-wired connections have obvious limitations for machines on the move. 5G is increasingly enabling industrial organizations to leverage powerful edge computing to bridge the traditional divide between digital and physical operations.

Actionable insights at the edge

The Deloitte study also found that early adopters of AI in manufacturing were driving an average of 10 different initiatives. These ranged from energy conservation and resource allocation to robotic process automation to collaborative engineering and digital twins. Due to supply chain disruptions and raw material shortages over the past two years, one area with a clear impact on results is quality detection and the detection of imperfect or otherwise defective products.

This real-time equipment monitoring requires cameras and nuanced AI analytics based on data models that can train the system to quickly “see” product defects and remove them from the production line, and alert engineers who can adjust specs or recalibrate machines to fix the root. cause. According to Deloitte and our own researchthese systems can reduce the shipment of defective products by up to 10 times.

Designing this type of edge computing solution requires cloud-like functionality on the camera sensor. While there are several ways to achieve this, the most effective approach is to use containerized software services that allow:

  • Efficient deployment of real-time image capture on the factory floor.
  • AI models that teach the AI ​​algorithm what to look for to generate insights.
  • A controller that oversees the operation, connects to a global manufacturing execution system (MES), and triggers an action based on that information.

Non-bypassable and compartmentalized functionality

As with any operationally critical system, security and safety issues must be addressed. One of the big challenges in system architecture is how to combine an organization’s IT networks with operational technology (OT) while ensuring worker safety, product quality, and system availability/reliability.

Any edge solution deployed in an industrial environment must be configured as a “least privilege model” in which applications only have access to the system resources they need to perform their function, and nothing more. These virtualized servers are effectively sequester the OT network from other parts of the organization’s network that are accessible to the Internet.

These permissions must be immutable to prevent reconfiguration (malicious or accidental) that exposes the “crown jewels” of a system. Think of the system as a house. All elements of the system are assigned to different “rooms” and each of them is locked, so that in the event that a specific area is compromised, access is limited.

There has been a significant increase in emphasis on protecting systems from attack. A more cautious approach is to recognize that no system is immune to attack; more effort must be made to recognize that a system has been breached and to return the system to a known good (safe) state.

Artificial intelligence is beginning to be applied here to learn the “normal” behavior of an embedded system and recommend when a significant change in the system should be assessed as a potential security risk.

Like other industrial infrastructures, manufacturing machines and software are often deployed for 10, 15 or 20 years or more. System designers must future-proof their solutions so that they are designed to remain impervious to any attack from any internal or external source for decades. Again, containerized software can allow segments of the software to be updated or upgraded without the need to replace the entire system.

Split-second decisions

In addition to complicating supply chain issues, the pandemic has also accelerated the need for new security measures designed to minimize human interaction and exposure. In response to this need, some manufacturers have deployed collaborative robots – “cobots” for short – to ensure their production facilities stay online to meet customer demand.

Cobots essentially interact with human workers in a shared workspace and are quite different from traditional industrial machines, which generally operate independently of human labour. As cobots find their place in the factory, it is paramount to ensure the safety of their human colleagues. This is a complex problem that is not easy to solve. A private 5G network, however, can provide powerful bandwidth for the cobot to act as an “edge computer” and make split-second decisions in response to the behavior of people next to them on the line.

The adoption of private 5G networks by today’s early adopters in manufacturing offers a glimpse into the factory of the future. As transformational and exciting as real-time capabilities and analytics can be, system designers must maintain the same rigor and attention to the safety and security of these networks as applied to core computing systems in the organization – if not more.

Comments are closed.