What is needed for the IoT to reach its potential? That was the theme of the opening day of Advantech’s online partner conference this week. Steve Rogerson reports.

Just over a year ago, I wrote here about how the predictions that the IoT would reach 50 billion connections by 2020 were so drastically wrong. The real figure was less than a fifth of that, and I talked with Stephen Carson of Ericsson, one of the companies that made the prediction, about why. At the time, he was saying things like these were not proper predictions and really Africa was to blame. Honestly, you can read it here.

This week, though, I listened to someone who had a more realistic grasp of why the forecasts were so far out. This was William Webb, a bit of an IoT guru and co-author of The Internet of Things Myth with Matt Hatton. He was speaking at Taiwanese firm Advantech’s online partner conference and reckoned the actual figure was around 8.5 billion. He still thought 50 billion would be hit but said it would more likely be between 2035 and 2040.

He said when the predictions were made, many in the industry had a too simplistic view of what would be needed to get devices to connect. As an example, he brought up connected litter bins.

This has been one of the big applications pushed by those in the smart city world. No longer would the bins be emptied at fixed times but sensors in the bins would say when they were full and an alert would go out for someone to come and empty them.

Now, that sounds great but let’s think it through. Local governments have systems in place to empty bins at fixed times. They have big trash collector lorries that have a set route to empty the bins. If they switch to the sensor system, what happens to the big lorries? They would be replaced by a smaller more agile fleet that would react on demand, something Webb called Uber for bin collecting. Behind that there would be an infrastructure to keep track of the alerts and notify drivers of which bins were full.

That does not come cheap. And most local government departments run to extremely tight budgets. Thus spending that kind of money not just to convert the bins but also to change what happens behind the scenes is probably quite low on the priority list. In most cases, the benefits just don’t justify the cost.

The second problem, and this is one most of us are familiar with, is how to make the connection. There are at the moment just too many ways to do this. Those building IoT devices cannot put all possible technologies on the same device, that would be too expensive, so they have to pick one, and they might pick the wrong one, or one that works now but not in a few years time. This problem has made many nervous about jumping into IoT too early. It is the old Betamax versus VHS story. By the way, do people still know what that means?

Interestingly, Webb gave as an example one area of the IoT that has taken off in a big way – wearable fitness trackers – and he said this was because these had just about all standardised on Bluetooth for making the connection to a smartphone.

The third area, and this is not that unlike the first, is that most companies are nervous about all the changes that have to be made to embrace the IoT fully. This even comes down to a change of culture and a change of personnel.

So that is why it went wrong, and much more believable than blaming Africa. But what about the future? Well, the whole opening day of the conference seemed full of optimism about this, and one reason is the merging of artificial intelligence (AI) with the IoT, in what is now being called AIoT. AI provides the means to get the most out of the vast amounts of data being generated, and see new connections and correlations. And deployed at the edge, it allows local decisions to be made.

“AI is a technology trend that we can’t ignore any more,” said Jerry O’Gorman, associate vice president of Advantech’s industrial IoT group. “In 2020, there was a significant growth in projects deploying edge AI.”

Linda Tsai, president of the industrial IoT group, agreed: “People talk about connecting millions of devices, but the real value is in the data they generate.”

Steen Graham, a guest speaker and general manager of Intel’s IoT group, said edge AI could transform industry. He gave vibration monitoring as an example. Analysing the vibration data from a machine could give early indications of when it might fail, allowing preventative maintenance.

The second reason for optimism is 5G, and what those fast speeds and low latencies can do for areas such as autonomous driving, even though many IoT applications work fine with much slower connections.

“5G is about to disrupt many industries,” said Dirk Finstel, CTO for Advantech in Europe.

Another guest speaker was Nicole Denil, Microsoft’s general manager for IoT in Asia. She said the industry was moving from connected assets to connected environments and in the future to connected ecosystems.

“Embrace IoT or be left behind,” she said.

Advantech’s CEO KC Liu summed it up by saying we had moved from the peak of over expectation in 2015 to seeing real growth and that by 2025 there would be a much improved IoT ecosystem, something that was needed if those early predictions were going to happen.

So, all very exciting, and I have learned from experience to listen carefully at Advantech events. It was, after all, at a physical Advantech conference many years ago that I first heard the phrase “internet of things” and thought that was silly and would never catch on.