June 26, 2020 | Denise Rael
Moving intelligence to the edge can lead to systems with better real-time performance, better power efficiency and enhanced security. But more intelligence requires more code,...
As the Internet of Things takes shape, companies are realizing that not everything can, or should be moved to the cloud. For many industrial applications, real-time operation and analog interfacing are crucial. While cloud-based analytics can bring new efficiencies and insights, computing at the edge of the IoT stays fundamental.
But if industrial computing needs the edge, the question then becomes – what does the industrial edge look like?
While the public thinks of the IoT as connected thermostats, smart lights, and intelligent appliances, the true value in the IoT will most likely come from the industrial market. For industries like manufacturing, supply chain and logistics, the ability to gather and use actionable data can reveal new insights and optimizations which reduce inventory costs, improve energy efficiencies, shorten time to market and much more.
But while the Industrial IoT (IIoT) can unlock a lot of value, the transition to cloud connectivity for industrial markets must take a different approach than the consumer side. A traditional cloud computing-based IoT model consists of a simple, low-power IoT sensor and actuator nodes which connect to cloud data centers. The cloud then gathers the data for analysis and sends back actions for the IoT nodes to perform.
While consumer devices like our phones, computers, and tablets can tolerate the latency of a trip to the cloud, this model doesn’t work for industrial applications. A missed or slow packet on an audio clip, web page or video is simply not a big deal, but for industrial processes, high-speed feedback and control are key. This difference in latency requirements is reflected in the networking standards used. While consumer devices use TCP/IP, industrial networks often use proprietary industrial Ethernet protocols to guarantee deterministic, low-latency communications between devices and systems.
Whether it’s an automated robot assembling a product or a smart valve in an oil and gas pipeline, industrial devices require real-time feedback and control. That means more processing ability is needed at the edge devices in the IIoT.
One can think of these IIoT edge nodes as having two sides: a cloud-facing side that transmits data to the enterprise, and an analog side facing the physical world, with the ability to take immediate action based on real-time data at the source.
In contrast to the consumer or enterprise IoT, the IIoT is therefore not defined by gateways and data centers – but rather by these edge devices, which need to interface with both the physical world and the cloud.
While many IoT nodes are sold as generic off-the-shelf hardware devices, real-world industrial applications often have specific requirements such as signal conditioning and Fieldbus connectivity, which can’t be completely met by a single piece of generic IoT hardware. This means multiple components must be combined, creating a larger, less efficient and more expensive device.
Even sensors, the simplest IoT node possible, often require specific analog and mixed-signal functionality to deal with non-linear input values and sensor calibration.
The scale of IIoT deployments – often in dozens, hundreds or thousands of devices – means that IIoT devices end up being price sensitive. Rather than combining generic hardware components to meet specific needs, device makers can save significant BOM costs with custom ASICs. Custom ASICs like Adesto’s SmartEdge Platform can create complete systems on a single chip which incorporate the analog front ends (AFEs), memory, security and the communication systems Industrial IoT edge devices need. By consolidating these components onto a single chip, space, energy consumption, and BOM costs are significantly reduced for the IIoT edge.