IoT roundup: Quantifying growth, medical imaging and 5G (have you heard of it?)

As 2020 approaches, we look at quantifying trends and ROI around the internet of things, an advance in medical imaging tech and bringing another type of network to the IoT game.

IoT > Internet of Things > network of connected devices
Avgust01 / Getty Images

There’s no shortage of glossy predictions about the global growth of IoT. These can be tough to place in context, given the wide range of methodologies involved, differences in exactly what the predictions are measuring and other variables.

Yet the Linux Foundation’s State of the Edge report gives an impressively ambitious forward look at the future of just one part of the IoT ecosystem, saying that the next decade will see more than $700 billion in total capital expenditure on edge computing systems, and that that tidal wave of spending will be driven in large part by IT infrastructure and data-center facilities.

Edge computing’s edge

That’s a startling number for those used to considering edge computing mostly as an enabler of digital transformation and IoT services, and a reminder that the edge is about both operational technology and IT. Having more hardware closer to users is a fundamental shift in the structure of the Internet, according to the report, and enables the use of more and more sophisticated location-aware applications.

Still, the foundation also predicts plenty of growth in traditional IoT sectors like manufacturing, with edge capex set to grow from $350 million this year to $3.55 billion over the course of the next six years, with a large amount of the growth driven by operational automation technology. Even the conservative healthcare industry will see vastly expanded usage, as the report predicts an expansion in the sector’s worldwide technological power usage from 2.1MW in 2019 to nearly 4.2GW in 2028.

Building better medical imaging

Speaking of medicine, researchers at MIT recently announced that they’ve come up with a clever new machine-learning system to help identify injuries and diseases in a programmatic way, opening up new possibilities in the field of medical IoT by generating better models for imaging analysis.

 An “atlas,” in medical imaging, is a template created by taking the average of a given patient population’s brain scan or chest x-ray, for example. The usual way to create one, according to MIT, is to simply average out all images of a given type for a given set of patients using a “lengthy, iterative optimization process.”

The issue is that the process can be complicated and lack precision if physicians are trying to create a template for a particular subgroup of patients if there isn’t enough raw data available. The new machine-learning system, however, can automatically figure out that certain variables correlate to the data in certain ways, making it possible to create useful atlases even if there aren’t a lot of specific examples for a given patient subgroup.

To put it another way – MIT has come up with a way to create “conditional” atlases. Even if a dataset doesn’t contain many brainscans of, say, 40-year-old men with a certain type of tumor, the system can apply what it knows from the dataset at large and come up with a fairly good estimation of what the baseline image should look like. That’s a potentially big step forward for some kinds of imaging-based diagnosis.

Have you heard the one about 5G?

Over in the networking space, ABI Research has discovered another facet of the modern world that will be heavily improved by 5G technology – heavy industry. According to a report issued by the market research firm, there will be 5.3 million 5G connections active in factory settings by 2026.

Most of the reason for that, according to ABI analyst Leo Gergs, is that 5G’s low-latency and high potential data rates are a great fit for the factory floor, providing a great way to implement something like a complex wireless sensor network or an augmented reality application for maintenance or monitoring.

“Furthermore, the technology opens up new production opportunities by enabling artificial intelligence applications to be integrated into manufacturing processes,” said Gergs in a statement.

This, of course, is going to be hard to check for the next little while, given that 5G isn’t really up and running yet outside of trial deployments in major urban areas and private implementations, but ABI points out that a critical part of 5G’s success will be catering to the enterprise market with specialist offerings.

“This should include moving away from selling connectivity as such and develop attractive pricing models for additional network capabilities,” Gergs noted.