IoT news of the week for May 15, 2020

AI could be the secret to better energy harvesting: For IoT to become truly ubiquitous we need to ditch both wires and batteries. Researchers at Newcastle University in the UK and the Uppsala University in Sweden have been working on using different dyes on photovoltaic cells to help them work more efficiently on indoor light. Which is great, but they are also using machine learning to help the device powered by the solar cell determine when it’s likely to have more access to energy so it can change its behavior. Instead of a binary sleep-wake cycle, future devices could use machine learning to anticipate typical lighting patterns and schedule energy-intensive events like a software update for times when they may have ample light. Researchers are already using ML to make batteries more efficient, so the concept is already out there. Of course, I’m not sure I want my devices shutting down in the middle of the day for a software update, but we’ll see. (IoTNow Transport)

How UPS is using the internet of things: Let’s talk about sensors at scale! TechRepublic sent a reporter to talk to Mike Allen, a transportation technology manager at UPS, and Juan Perez, the company’s CIO, and wrote a five-part series on how the logistics company is deploying sensors and changing its operations. More than 100,000 trucks are outfitted with sensors and GPS at UPS, leading to millions of messages a week, some 50-60 million of which are simply location data. The goal is to get sensors that measure the volume of packages each truck and electronic door locks for the trailers. I wish the series spent more time diving into how these sensors are powered, maintained, and their data ingested into buckets for real-time, short-term, and long-term use. But in another article in the series, we do get to read about drone delivery. (TechRepublic)

Wink’s latest SNAFU: The Wink smart home platform abruptly went to a subscription model last week, giving users a week to pay a $4.99 subscription fee or see their devices stop working. Based on “feedback,” Wink’s management has now given users another week (until May 20) to decide whether or not to upgrade to a subscription fee. At the same time, I just received a notification from my connected garage door opener letting me know that it was no longer going to let me control my garage door using Wink. Indeed, I can still see my Chamberlain MyQ status in the Wink app, but I can no longer control it. I’m not sure if this is an effort by Chamberlain to stop supporting a failing platform, or if it’s a cost-cutting measure of its own. Chamberlain has had its own issues in the past with charging users for accessing its device through other platforms. We’ll keep an eye on this. — Stacey Higginbotham

Trend Micro anticipates the rise of OT-specific malware: Right now, much of the perceived risk in connected manufacturing environments is focused on malware infecting IT networks and, in the process, stealing data or causing other problems. But as the IT and OT worlds converge, Trend Micro anticipates more threats that are designed to take advantage of these proprietary OT networks. To mitigate the risks, it suggests putting more integrity checks in place on OT networks (in other words, not trusting information on these networks as much as before) and developing new detection tools designed to find malicious OT code. The other takeaway is that these formerly air-gapped networks might have a smaller or non-existent gap now thanks to industrial IoT. So the OT world needs to spend more time (and dollars) on security. (Trend Micro)

More money for machine learning at the edge: SiMa.ai is the latest chip startup to get funding for edge-based machine learning, raising $30 million in a round led by Dell Technologies Capital with help from Amplify Partners, Wing Venture Capital, and +ND Capital. The company is designing a more efficient chip to handle machine learning, but it sounds like it won’t be ready for sampling until 2021. This is a huge and growing sector of the internet of things as companies recognize that bandwidth, privacy, and latency force more machine learning to happen close to the sensors as opposed to sending the data to the cloud for processing. I write about one of these startups twice a month, so I’m going to give this one a moment before getting too excited. (TechCrunch)

Open-source RISC-V chips are available for all: The CHIPS Alliance, which is an offshoot of the Linux Foundation trying to make open-source instruction sets and silicon designs available, has tweaked two RISC-V-based cores to make them more appropriate for use by companies seeking to use them for embedded and high-performance designs. The Alliance has validated the SweRV Core™ EH2 for high-performance jobs such as machine learning and the SweRV Core EL2 for low-power, embedded jobs.  The validation included testing how the cores will work with various design and testing tools. (CHIPS Alliance)

A nice update on IoT in practice: Over at Network World, the editor has created an essay that covers some broad trends and topics as a way to showcases a dozen or so stories the publication has written about the use of IoT. Check it out for an easily digested overview and links to some case studies you may not have seen. (Network World)

More thoughts from me on privacy and CVOID-19 tracking: In this month’s column at IEEE Spectrum I discuss how track-and-trace efforts can safeguard user privacy while still meeting public health goals. It’s probably familiar to those who read the newsletter, but I’ll leave it here for those who are into this topic. (IEEE Spectrum)

I reviewed the newest August smart lock with Wi-Fi: Click on through to see what I think. (StaceyonIoT)