AI equipment and better IoT networking
If there is ever a technology that did not live up to the initial hype, it would have to be the Internet of Things (IoT). After more than a decade of unfulfilled promises that promised how businesses, cities and organisations will connect office lighting systems, 30-year-old production equipment, parking spaces and, well, all to the Internet, there’s not much to show for joint efforts.
If there’s a real possibility of shifting the existing state of IoT, Microsoft unveiled an exciting host of IoT-related innovations at this year’s Ignite conference that showcases the road forwards, like a family of Arm-powered smart sensor appliances. In addition, the organisation revealed the general release of a range of essential, but historically ignored technologies that were likely to require them to take a few measures backwards before they could move anything forwards.
But I get ahead of myself.
From the beginning, the ideas underlying IoT and the challenges expected to make it useful were… optimistic. The measures involved in getting the sensors connected to all of these things—or tuning into the current sensors they had—networking all of them, gathering data reams, and then finding out what it all meant is hard—true hard. The idea that it also involved having to bring various aspects of a company that don’t normally take care of each other-namely IT people and OT (operational technology) people working together only added insult to injury.
One of the greatest obstacles is that the vast majority of organisations were-and still are-not yet to the point that it makes sense to do an IoT initiative.
One of the greatest problems, though, is that the vast majority of companies were-and still are-not quite to the point where it makes sense to carry out an IoT initiative, let alone one that will result in a total transformation of their business (as many were falsely promised). Logically, for example, an enterprise would need to know precisely what it had on its numerous networks before it embarked on an attempt to collect useful data from these multiple devices.
Turns out, since businesses, municipalities and those participating in IoT are usually older organisations with a lot of older equipment, even the basic criteria wasn’t as easy to fulfil as one would expect. In recognition of that, last year, Microsoft unveiled Azure Defender for IoT, exploiting the unique features of the technologies they purchased from a firm named CyberX that could instantly discover even older computers using what they call “agentless” technology.
What this means is that even older ‘brown field’ devices lack modern software agents that can report their capabilities back to a software management system and still rely on advanced industrial communication protocols and their capabilities can still be monitored.
Taken together, all these modern technologies and resources offer the crucial look backwards required to allow conventional organisations to even imagine beginning a big IoT initiative.
More specifically, like its laptop equivalent, Azure Defender will instantly instal security patches and check for misconfiguration errors, all of which have proved to be massive (and costly) security holes for IoT-related projects.
A few tweaks this year are that Microsoft is introducing additional features to find devices on nested hub edge networks, such as those utilising a protocol called ISA-95, that are usually “invisible” to other network-based management tools. The business also addressed a new dedicated piece of hardware named Azure Defender for IoT Sensor that connects and helps businesses to track these older, unmanaged machines. Taken together, all these modern technologies and resources offer the crucial look backwards required to allow conventional organisations to even imagine beginning a big IoT initiative. While some can see them as too plain, Microsoft has found that there are a number of very basic steps that most organisations need to take to get to the starting line.
Beyond linking and protecting all devices, another core feature of IoT is creating practical information from all data that can be created by a network of connected devices. Again, it was much harder for most organisations to be able to achieve this than we realised. To meet this need, Microsoft unveiled Azure Percept, a set of AI-accelerated edge computing devices and accompanying Ignite app platforms this year. Azure Percept consists of a series of pre-built appliances, computer vision and audio-based machine learning models, and a basic development environment called Azure Percept Studio.
Azure Percept Studio is one of the large low-code, non-code-based applications unveiled by Microsoft at this event (including Excel formula inspired by PowerFX and the incorporation of PowerAutomate automation/macro platform directly into Windows 10), helping people to create sophisticated machine learning models using either the camera sensor data from the Azure Percept Vision module or the Azure Percept Vision module.
Since the models that Percept Studio produces are containerized, cloud-based code pieces, they can be quickly ported to different platforms, further built in more sophisticated programming tools, and moved around multiple cloud or edge-based computing settings. In addition, the Percept platform gives end-users access to far more advanced Azure AI Cognitive Services and Azure AI Machine Learning models.
This first version of three Percept hardware modules is all designed by Asus, but Microsoft plans to provide additional hardware partners for the Percept platform later this year. Initial devices include a Development Platform Package built on an iMX8mq SoC Arm-based NXP with CPU, WiFi, Bluetooth, and Ethernet support, as well as a TPM 2.0 module to allow trust root hardware.
In addition, there is Percept Vision, which is designed around the Intel Movidius machine vision processor, and Percept Audio, which comprises a far-field array of four microphones, support for custom keyword and hardware-accelerated AI processing. The bundle of the development kit module and the Percept Vision sensor will be priced at $349 and the Percept Audio sensor will be priced at $79. All three systems feature an industrial design that leverages the 80/20 profile standard for aluminium tubes, making it very simple to instal or build various parts into specific devices, like robots.
Although the Percept Platform is fascinating on its own, it once again exemplifies how Microsoft is extending its expertise to the many simple problems that prevent most conventional businesses from discussing and tackling the IoT project head-on.
At a high level, it also seems to be part of the newfound emphasis on the practicality and utility that Microsoft is showing at Ignite this year. Several of the company’s announcements-including stuff including Uniform Print support for Windows and Microsoft 365-are unlikely to be really interesting. Yet, like any of the latest IoT and smart edge efforts, they are really useful to a lot of people.
Plus, they also demonstrate the business thinking about very basic, but often ignored, obstacles that have stopped many companies from implementing any of these innovations more generally. I hope this is the beginning of a greater age in which even basic technological needs are not taken for granted and remote helping hands are provided to all individuals and organisations that really need them.