The edge, and edge computing, figure prominently in a lot of internet of things (IoT) discussions right now. But where — and what — the edge is very quickly gets complicated. Telecom operators understand the edge of their network. Sellers of servers and gateways know what they’d like their customers to fill the edge with. Beneficiaries of IoT accept that data, value, and control often move between the center, the edge, and various intermediate points. But the terminology doesn’t really help any of them speak to one another, because they all imbue these simple — and apparently obvious — terms with lots of context and baggage.

A wind turbine in Canada
Source: Paul Miller

In my latest report, “Make Room For The Autonomous Edge In Your IoT Strategy,” I suggest four distinct ways to think about the edge. I also look at Forrester data to explore how our clients currently think about the challenges and opportunities associated with working at the edge.

To segment our discussion of edge, I propose separating it into four (often overlapping) segments:

  • The sensing edge. Small, cheap, low-power sensors propagate through the environment to measure everything from light, temperature, wind, or vibration to the opening of doors. This class of edge sensor lacks the ability to interpret or act upon the data it collects and must rely on intelligence elsewhere to transform the stream of ones and zeroes into actionable insight.
  • The actuating (or moving) edge. Day after day, small, cheap, low-power devices on the actuating edge close valves, activate pumps, angle turbines into the wind, switch trains from one track to another, or lock doors, all under the control of some remote intelligence, either human or virtual.
  • The stateful edge. Extreme operating conditions and intermittent communications with far-off data centers mean that IoT-enabled solutions must be designed to cope with the unexpected. These devices may require local storage or computational ability to preserve data that cannot transmit immediately or to make decisions when the remote control room is unable to send instructions.
  • The autonomous edge. Autonomous edge is comparable to the concept of edge computing. On the autonomous edge, cameras watch station platforms or industrial processes, and locally deployed algorithms make the call to redirect passengers, alert the police, or flag a set of products for quality control.

What do you think? Does that help or hinder the conversation, particularly when (for example) networking people, infrastructure people, and business people need to communicate?

As always, Forrester clients can schedule an inquiry call to ask about this. Anyone can schedule a briefing to tell me what they’re doing.