If you want to solve a business problem using a computer, you have to connect to it. The furthest point at which you can connect the “edge” has always been a major frontier of computing. In the 1950s, the edge was where you were close enough to feed punch cards into a computer.

As computers got smaller and cheaper, networking expanded access, and the edge spread beyond centralized computing hubs. Today, mobile connectivity and the cloud have pushed the edge further. Consumers and businesses have anytime access to virtually unlimited computing power. We have the ability to connect billions of small devices, creating an Internet of Things. And we’re starting to see IoT scenarios that demand bringing the computing power to the problem.

That’s where the autonomous edge comes in. In Make Room for The Autonomous Edge in Your IoT Strategy, Forrester defines the autonomous edge as “A family of technologies that distributes application data and services where they can best optimize outcomes in a growing set of connected assets. It includes edge infrastructure and edge analytics software.”

In other words, it brings intelligence to where the problem is. The decisions and actions in autonomous edge computing happen out in the world.

IoT phone home

One challenge is connectivity. Anyone who has had a streaming movie interrupted just before the big plot twist knows that internet access is not guaranteed.

Now, imagine that instead of playing a movie, you need a device to decide whether an oil pump on the Russian steppes is about to explode, or whether a hidden flaw in a high-speed production line is going to create massive waste.

In the first case, you might not have a high-speed internet connection available at all—and lives are at stake. In the second, milliseconds of time could translate to millions of dollars. These are problems ideally suited for autonomous edge solutions.

Autonomy in action: a few examples

Whether it’s in self-driving vehicles, mixed reality, or smart buildings, the autonomous edge has incredible potential to transform our lives for the better. For businesses, it can deliver a wide range of benefits, as the Forrester report shows:

• Handling present and future AI demands
• Avoiding network latency and allowing faster responses
• Reducing the need for expensive network connectivity at remote locations
• Handling pre-processing for an ever-growing number of IoT devices

Companies are already realizing these benefits with innovative autonomous edge IoT strategies. Take Schneider Electric, for example. Its Realift Rod Pump Control allows oil and gas companies to monitor and configure pump settings and operations remotely. Schneider wanted to push this capability further.

“If you look at most of the controllers that exist in the market today, they are reactive, looking at what is happening now and responding accordingly. We want to be proactive and include predictive analytics at the edge. It’s a real game changer.” – Helenio Gilabert, Director for SCADA and Telemetry at Schneider Electric.

Using Azure Machine Learning and Azure IoT Edge, the company is doing just that. Running predictive models right on the controller, Realift can sense anomalies that show impending problems, change settings, or even shut the pump down to prevent damage.

Processing video streams is another powerful application of autonomous edge. Intelligent algorithms can analyze video and take actions based on what they “see” including counting open parking spots, finding gaps on retail shelves, detecting manufacturing defects, and more.

The problem is that raw video streams are big. Pushing them to the cloud in full can be slow and expensive—especially if you’re talking hundreds or thousands of cameras, or remote areas where bandwidth costs add up quickly.

That’s why Microsoft and Nvidia partnered on a new approach. An edge gateway and advanced camera-stream processing analyze videos locally. The solution gleans the important data and sends it to the cloud as needed rather than sending the videos themselves. This delivers real-time performance and reduces compute costs.

Overcoming the four biggest challenges to a smarter edge

Although the potential business value is huge, there are some unique challenges to overcome. According to the Forrester report, the top four barriers identified in a survey of over 1,900 global telecommunications decision makers include security, organizational barriers, device management, and cost. Let’s take a look at each one in turn.

1. Security

Autonomous edge devices have attack surfaces that are similar to traditional computers. Plus, they run in public places. Extra security measures are critical to protect against the real threat of malware and other attacks. Fortunately, technology is keeping pace. For example, Azure Sphere combines security at the hardware, software, and cloud levels to enable new levels of trust with intelligent IoT devices.

2. Organizational barriers

When it comes to the autonomous edge, not all the challenges are technical. Operational silos can also block progress. Because autonomous edge devices share many similarities with computers, the IT organization may have the skills to manage them, yet lack the authority to do so. Leveraging your information services department’s skills can help reduce the cost and complexity of edge projects.

3. Device management

Managing a mobile phone that’s always connected to the internet is one thing. Managing a well pump that connects once a week is another. Thinking through the strategy is important. How often do you need to update the OS or software? Will you push updates over the air, or will technicians install them manually? How do you check the status of the edge device separately from the telemetry it sends to the cloud? It makes sense to start with purpose-built IoT monitoring and management solutions.

4. Cost

The smarter the device, the more expensive it’s likely to be, so weigh the value of autonomy. A simpler approach may be better. Some use cases just need a device that can store data and computing state when disconnected. Some need to actuate a physical object, but don’t require the IoT device to make the decision about when to do so. All the same, keep Moore’s law in mind: What may not be cost effective today may be within reach next quarter.

Find your edge with Azure

At Microsoft, we’ve staked our success on innovation that spans cloud and edge. We can help you get started quickly with solutions such as Azure IoT Edge, a fully managed service that allows you to run IoT workloads at the edge so your devices spend less time communicating with the cloud, react more quickly to local changes, and operate reliably even in extended offline periods.

Learn more about Azure IoT Edge, and how you can use the autonomous edge to your advantage.