IoT news of the week for June 15, 2018

Resin.io gets $5 million: Startup Resin.io, which puts containers on devices running ARM-based chips, has scored $5 million in funding. I wrote about the company a few months back, explaining why what it was doing was important for the internet of things. It’s basically making it easier to manage and write software to a large swath of connected devices. That way, developers used to the ease of deploying software in the cloud can turn their skills to the internet of things. Making developers’ lives easier is a good way to get money. (GeekWire)

Amazon launched a $250 camera for the edge: Amazon is now shipping a camera that helps developers build computer vision applications. Amazon’s DeepLens camera ships with access to a variety of AWS services, the idea being that developers can run computer vision models at the edge. The camera comes with several models already installed, such as face detection, cat and dog detection, and even things such as head pose determination. So using the camera, a developer could build an application that recognizes a person in a restricted area and then triggers an alarm. Or in one of my favorite ideas, recognizes a dog on the couch and then triggers a speaker to tell the animal to get off. (Engadget)

Hackers are getting pretty bold on your IoT devices: Hackers are aggressively targeting your connected devices. Allot has released a report on the state of telco network security that shows how far hackers will go to get their malware onto a connected device and keep that malware running. Allot set up honeypots designed to resemble unsecured IoT devices and waited to see what would happen. The report said immediate, successful attacks on the devices peaked at a rate of 1,000 per hour. Based on the aggressive speed of attacks on the honeypots, Allot calculated that a connected device will get infected within 42.5 seconds, on average. What struck me was that once on the device, hackers quickly moved to claim it, checking to see if it was real, trying to obliterate other malware on the device, and locking it down for the hackers’ use. That use might be mining Bitcoin or playing a role in a botnet attack. Before we all panic, it’s worth noting that what Allot calls an IoT device is generally something running Linux that has enough computing power to be worthwhile. These are routers and set-top boxes, not light bulbs. (Allot)

​Industrial sensors need industrial batteries: This article is why I love trade publications. Where else would I find a deep dive into the properties of seven different batteries as it relates to the industrial internet of things? The article shares all of that and begs the industry to come up with even more battery tech aimed at smart manufacturing and industry. If nerding out on deep tech is your jam, this is for you. (SensorMag)

Do we design buildings for people or machines? With the news this week that Microsoft is working on technology for retailers to enable checkout-free store like Amazon Go, this story on future architecture rung a chord. Microsoft hopes to create a video-based shopping experience that tracks what people put in their carts and then lets them leave the store without checking out; the tracking tech would automatically deduct the cost of those items from the customers’ stored payment credentials. The story on architecture laments the demise of buildings designed for people. It focuses on banks, airports, and other venues where people interact more with machines. A significant chunk of it talks about self-checkout centers. The story felt a bit overwrought, but when juxtaposed against the Microsoft plans and the Amazon Go concept, it also felt a bit dated. While today we might design for big bulky machines to accept our cards and for self-scanning of items or whatnot, we aren’t far from a future where video and biometric credentials make bulky self-checkout machines or airport kiosk obsolete. At that point, what kind of spaces will we need? Likely ones where we can go to get a little privacy. (The Atlantic)

A broad overview of IoT in food safety: This article is pretty basic, but it does a decent job of explaining the places where sensors and data analytics can be used to make our food supply chain safer. While I’ve looked at individual elements such as cold-chain tracking (making sure food stays cold during shipping and in storage), this piece talks about sensors that can track pathogens in warehouses and other venues. The story is written by the company that provides pathogen-detecting sensors, but given the constant barrage of news about food-borne illness, I’m interested in the idea. (IoT for All)

A detailed dive into AI attacks: While we’re wringing our hands over the number of malware and ransomware attacks hitting the internet of things, there are people wondering how to take hacking to the next level. These folks are thinking about how we fool computer vision, break statistical models so AI automation fails, and steal models so they can game them or reuse them for their own profits. The article is important, because people seem to think that computers are infallible, forgetting that garbage in equals garbage out. Many of these attacks are basically attempts to feed computers garbage so they can’t do the job they are supposed to do. But right now we’re so in awe of AI we forget that, and thus, don’t guard against it. After reading this article by a computer scientist employed by Google, you may not be as concerned over AI as you once were. Or maybe you’ll be more concerned about our faith in it. (Elie)

Now this is longevity for IoT! This week NXP said it would support its current Kinetis line of microcontrollers for the long haul, promising that it would keep the line for 15 years. This promise isn’t related to security updates, but rather a plan to make sure the products are stocked and stay relatively unchanged so when companies design a product around them, they don’t have to redesign them for years. This is similar to a recent promise by Qualcomm to extend the longevity of its Snapdragon chips for embedded systems to 7 and 10 years. This is a necessary evil for an industry where replacement cycles are closer to decades, not every few years. So when looking at your shiny new connected device consider the silicon inside. Is it set for the long haul?  (NXP)

Did you know Intel processes more than a 5 billion data points a day in its factories? The chipmaker laid out a white paper talking about how it deals with data. Yes it’s selling Intel products, but as a manufacturer that makes billions of chips a year, it’s also practicing what it preaches and can offer useful insights to others. (Intel)