IoT news of the week for April 3, 2020

Carriers and clouds make for an uneasy alliance: This week, two announcements came out that are worth discussing. Nokia announced a new service, called AVA Cognitive 5G, that helps telcos manage some of the challenges associated with building 5G networks, and Microsoft announced new edge services called Azure Edge Zones. In my call with Nokia, we discussed the challenge that carriers and cloud providers face when it comes to delivering low-latency services using 5G networks. When people were using telco networks to talk to other people, a bit of latency was fine, but as we put machines talking to machines on these networks, we need sub-millisecond latency that can be hard to deliver without having both the cloud services and the network tightly integrated. Microsoft’s Edge Zones are an example of this, with Microsoft putting elements of Azure on carrier networks closer to their customers, or even shipping dedicated boxes to their customers to run on-premise. (Microsoft)

NXP makes it easy to put Microsoft’s Azure RTOS on its chips: Did you know that Microsoft has its own Azure real-time operating system for embedded devices? Much like Amazon — which had hired the creator of the popular FreeRTOS real-time operating system and subsequently linked FreeRTOS to the Amazon Web Services cloud — Microsoft also has an RTOS, which it acquired last year when it purchased Express Logic. And now, for customers who want to use the Microsoft RTOS but don’t want to spend a lot of time negotiating with Microsoft, they can simply purchase some of NXP’s silicon and have the SDK associated with Microsoft’s Azure RTOS already on the chip ready for use. This will help make it easier to get started with an IoT build and require less effort when it comes to integrating the RTOS for their hardware — and later tying that RTOS to a cloud. (NXP Semiconductor)

The internet is NOT breaking, y’all, and here’s proof: Christian Koch, who writes an excellent newsletter about internet infrastructure, has published a series of slides detailing traffic growth during the COVID-19 pandemic. He’s offering clear proof that while traffic is increasing, and some service providers are trying to take advantage of the situation, for the most part the internet is not breaking under the pressure of people working, schooling, and staying at home. (Christan Koch)

The difference between making decisions and making predictions: I get super frustrated with the people who talk about data being the new oil, because unless you turn data into some kind of insight it’s just a bunch of bits and bytes costing you money in a cloud database somewhere. Even more challenging to talk about with people is what, after we have successful algorithms that can make predictions, comes next. So far, one of the more ironic things about adding automation is that employers find they need staff to make higher-level decisions about what to do when an algorithm lets them know something is about to fail or is slowing down. Making predictions only meets half the challenge; someone still has to decide what to do with that information. This guest columnist in The Wall Street Journal did a great job explaining what it means to bring AI into the decision-making process, and how we need to review our algorithms to ensure those decisions make sense. (WSJ)

A good model for remote medical monitoring: When I first saw from an Israeli medical device company a connected device designed to monitor maternal and fetal heart rate, I worried that it would be another example of digital snake oil, a device that advertises some medical capability without any real proof. But this is remote monitoring done well. The NUVO device must be prescribed by a doctor and is worn during the third trimester as a potential replacement for the almost weekly doctor appointments that occur late in pregnancy. It lets the doctor set a regular appointment to monitor the data from the device, which in turn lets doctors adopt the NUVO device within the current standards of care. One of the challenges with wearable device-derived data is that doctors are worried that somewhere in the data they will miss something important; scheduling a remote “appointment” to monitor the heart rates get around this. (Medgadget)

Miss those Amazon Dash buttons? Seeed Studios wants to help: Seeed Studios has built a sub-$20 button that you can use to connect directly to the Amazon Web Services IoT 1-Click service. The button will let developers trigger an AWS Lambda function to execute an action. Such an action may be a request to order more printer paper or to call customer support. You can set the action up yourself and then label the button accordingly. (CNX Software)

Washington passes a facial recognition law: The state of Washington has passed the nation’s first law governing the use of facial recognition. Essentially written by Microsoft, the law allows the government to implement facial recognition only after filing a public notice and holding meetings with the community. It also allows for law enforcement to use the technology, but says they must have a warrant. The law does carve out an exemption to the warrant for “exigent circumstances.” Unlike several cities that banned facial recognition outright, this law sets parameters around how the government can use the technology. As expected, some are happy with the new law, while others are concerned about how it will impact vulnerable communities. I am a fan of warrants for use of facial recognition in law enforcement and I do like the fact that this law makes a point of allowing third parties to audit any facial recognition tech deployed by the government. But I am worried about the broad exemption for exigent circumstances. (Geekwire)

COVID-19 takes down NTT DoCoMo’s NB-IoT network: I wasn’t expecting this, but Japan’s NTT DoCoMo has shut down its nascent NB-IoT network, saying it needs to “concentrate management resources.” The operators will still keep its LTE-Cat1 M network running, but the super low power and low data rates promised by NB-IoT will be stalled for a while. Basically the low-priced data plans that NB-IoT promises don’t justify the expense of upgrading, which has me wondering if we’ll see other carriers — especially in the U.S., which also was a late adopter to NB-IoT — pull back on their network rollouts. (Fierce Wireless)

Thingstream’s MQTT expertise acquired by u-Blox: Swiss location company u-Blox has acquired Thingstream, a company that optimizes MQTT messaging protocols popular in the IoT. The deal is a small one, with u-Blox paying just 10 million CHF ($10.27 million) for the company, but worth mentioning as the style of messaging popularized by MQTT is becoming a standard for delivering data to machines and web services on a constant basis. Companies are embracing messaging over the more active API calls because they can be easier to set up and manage over time. This brings u-Blox deeper into the IoT. (u-Blox)

A deep dive on functions-as-a-service: For the companies out there building IoT services, serverless computing has emerged as an essential element used to reduce the costs associated with connected devices talking to the cloud. With serverless computing, a cloud provider only spins up a server when a device sends a message, as opposed to keeping the server running all the time. This cuts costs. But there are subtleties in every form of functional computing, and Richard Seroter really dives into them in this blog post. He spent time deploying jobs to several different platforms to explore the difference between those designed to connect computers, those designed to expand on a host platform, and those designed to deliver complete web apps. Architecture nerds are going to want to read this. (Richard Seroter)

Interested in sponsoring this newsletter and the IoT Podcast? Get more information and contact us here.