In the IoT, billions of devices will connect to the cloud, but they need to be able to process their own real-time data at the edge.
I’ve often thought the most valuable interactions happen with the people at the edge of our networks. The people we meet serendipitously, through our more distant contacts. It’s here, on the edge, where the sparks of creativity really fly.
Recently, I’ve been putting this theory to the test by taking the time to meet face-to-face with people I might more usually only connect with by email or LinkedIn.
It’s a refreshing experience. One of the many benefits is frequent exposure to new ideas and new ways of thinking from people who view the world through a different lens. There are other benefits, such as getting to see the new musical “Bat Out of Hell” in the company of an American lawyer. But I digress ...
It’s not just people who benefit from networking at the edge. Computers do, too, and there’s an interesting analogy to be drawn here with the emerging importance of edge computing. This is where processing and data are placed at the edge of our networks to have the maximum effect.
Let me explain.
Over the last decade or so, we’ve seen processing and data increasingly centralized in the cloud. This has been driven partly by the cost-effectiveness and scalability of cloud computing and partly by the growth of big data.
Amazon Alexa is an excellent example of how this works. Voice commands are picked up by an Amazon Echo device, converted from speech to text and fired off to the cloud where natural language processing (NLP) and clever software are used to interpret and fulfill your requests. The results are served back to your Echo in less than half a second. Very little processing takes place on the Echo, and very little data is stored there; all the heavy lifting is done centrally in the cloud.
See also: The Big Lesson From Amazon-Whole Foods
This model works well if the edge device (the Echo) is always connected to the cloud via the internet, the arrival rate of new data (your voice commands) is relatively low and the response time is not critical (we’re happy to wait half a second for Ed Sheeran to start his next song).
But it doesn’t work so well if the edge device is not always connected, if the volume of real-time data streaming into the device is huge or if an instant response is needed.
Imagine you’re being driven to the theater by your AI-controlled smart car equipped with hundreds of sensors gathering real-time data from every direction. If the sensors detect a child running out in front of the car, there’s no point firing that data off to the cloud for processing. It has to be processed and acted on instantly and locally by the car itself.
There are many, many more examples where the edge devices (cars, traffic lights, fitness bracelets, microwaves, safety critical sensors on assembly lines… in fact, very many of the billions of devices that will be connected to the internet of things over the coming years) will need the ability to process their own real-time data.
These edge computing devices will still connect to the cloud, but the location of the processing and the data will vary according to need — in the cloud for asynchronous machine learning insights and improvements; at the edge for real-time processing of real-time data streams to determine real-time actions.
Developing the hardware and software for these devices will require new ways of thinking. It’s not about big data; it’s about small, fast data. And I’m sure we’re going to see dramatic improvements in battery efficiency, data storage and processing capability of these intelligent edge computing devices.
The Internet of Things is actually going to become the Internet of Small Powerful Intelligent Things (although I doubt that acronym will catch on).
See also: Major Opportunities in Microinsurance
Most interestingly of all, though, from a cultural and business perspective, is the innovation this edge computing will enable, such as:
- The insurance industry will be able to offer better deals and new types of policies driven by the intelligence embedded in the insured assets.
- The health industry will be able to provide preventative care supported by intelligent wearables monitoring everything from activity to blood sugar levels.
- The entertainment industry will be able to deliver interactive content without those annoying buffers and whirling circles.
And, who knows, maybe edge computing will also help us communicate more effectively with each other. Because spending time at the edge of our networks, as I have been discovering, is where the sparks of creativity really fly. Like the musical “Bat Out of Hell,” it’s one experience I can thoroughly recommend!