IoT news of the week for Oct. 4,2019

NXP has a gigahertz-powered microcontroller! NXP has announced a microcontroller — traditionally a super-low-powered processor designed for embedded computing needs — that delivers 1 gigahertz of processing power. This is what a top-of-the-line computer ran back in 2000. Historically you haven’t needed a ton of juice to run a microwave clock or a washing machine, but thanks to connectivity and the demands of smart devices, a few megahertz are not always enough. NXP is aiming the new MCU at products that will run machine learning inference at the edge, trying to match incoming data to an existing model. This is a dual-core MCU designed for industrial markets and frankly, I’m shook at the idea of an MCU with this much performance. (NXP Semiconductors)

MIT Researchers create a distributed location network for IoT: It’s hard to find the precise location of a wireless device in environments where GPS signals don’t penetrate well (inside buildings or in urban areas with tall skyscrapers). It’s also difficult to pinpoint a location where many connected devices create a lot of wireless noise that can stop devices from calculating the strength of nearby wireless signals to determine where something is. But researchers from MIT and other universities have created a machine-learning algorithm that can alleviate some of the problems associated with getting an accurate location in both situations using machine learning and a variety of data from nearby devices. The researchers call it “soft information” and it sounds like the calculations performed by the algorithm are akin to when I walk around without my glasses and infer from a variety of blurry data where my glasses are. I’m looking at shapes, thinking of last known location, pulling from history and maybe even can ask my husband. It sounds like these algorithms are calculating their location based on a similar variety of hazy factors. Either way, this is important research because, with a world full of sensors, we’re not going to be abe able to tell the computers exactly where they are. Those connected devices need to do that themselves. (MIT)

Here’s an outdoor camera that does local image recognition: Worried about connecting your home security camera to the cloud? Plenty of people are, but they also want features that traditionally require a cloud connection, such as image recognition. Netatmo’s Welcome camera has offered offline facial recognition for years, but few other companies have. However, Chinese company SimShine built an outdoor-rated home security camera with AI-based features that will compare with those from companies such as Nest or Ring. But the Alloy won’t require an Internet connection or subscription. The company has just raised $8 million to bring the device to market. (VentureBeat)

ReFirm Labs gets funding for IoT security: Like I said in the lead story, if you have an IoT security angle, you can get investor attention. ReFirm Labs has a product to asses IoT device firmware security, and has already managed to get several customers. So far, telcos and cable companies seem most interested in ensuring that the devices they install for customers are secure and stay that way. (CRN)

Train an AI to handle your privacy: A service called Guard wants to help users protect their privacy by analyzing the language in apps’ privacy policies as well as their prior history of protecting user security. The service uses machine learning to read privacy policies and also tracks hacks and data breaches to make its judgments. As an added bonus, you can take a quiz to help train the models used by Guard to figure out the most privacy-preserving language inside terms of service. Of course, this implies that people actually understand what options are best for user privacy. I can see lawyers trying to adjust their language to make it sound more protective as a way to fool the algorithm over time. (Vox)

The IKEA smart blinds review: The good news is The Verge has tested the IKEA Fyrtur connected blinds, and likes them. The bad news is that the blinds are still not for sale in the U.S., despite IKEA giving an availability date of Oct. 1. This is a product people are avidly waiting for, so I understand the frustration. Inexpensive smart shades are something people regularly ask us about, and as someone who spent $700 at my old house to connect my $10,000 shades, I can tell you it’s completely worth it. The reviewer says that setting up the Fyrtur blinds is a pain, but once that’s done, they reportedly work well. The reviewer also wanted a sunrise/sunset trigger as opposed to merely relying on a set time. I get that. (The Verge)

Are you a software company, or something else? Two of my favorite writers that cover capital markets, Dan Primack and Fred Wilson, are having an ongoing conversation about what current valuations for recently public companies signal. For startups or even later-stage private companies, this is important, because it lets them know where their valuations should be and how to structure their business if they are early enough in that process. Wilson comes down on the idea that if you can deliver software margins of more than 50%, your company can be valued like a software company. However, for companies that are below that threshold — certain companies with high licensing costs or hardware in the mix, for example — then you’re “something else.” Several IoT companies will likely fall into the “something else” category, which means that investors are still trying to figure out what the heck those companies should be valued at. I can’t wait to see this conversation evolve. (AVC)

Look, it’s smart home porn! No, there’s no actual porn here. But this is a beautiful home outfitted with a home automation system built around the Sabbath. One of the unsung benefits of a good automation system is that it can be automated to respect the Sabbath. Depending on your level of observance, you can set the house to do the work you can’t or you can turn it off so nothing works on at all. It’s also a fun look at a $250,000 smart home system. Egads! (WSJ)

The tariffs are coming! Wyze is the latest company to inform their customers about how proposed tariffs will affect their pricing. So far, Wyze says it is working with its manufacturing partners to avoid passing on what it calls “a shiny new 15% tax we have to pay Uncle Sam.” Wyze is known for its high-quality, low-priced gear. But in its tariff email to customers, it notes that while it makes a profit on its products, it doesn’t make enough of one to cover the tariff increase. So it is working with manufacturers to find ways to be more efficient while trying to avoid raising prices. Smart oven maker June has also had to raise its prices to make up for increased goods prices because of tariffs, and I’ve talked to executives at larger smart home device companies who are upset about the tariffs but are eating the cost for now. I do expect prices to go up on goods from smaller vendors, and expect to see prices for smart home devices fall more slowly as manufacturers try to deal with these new costs.

Kevin is not sold on Amazon’s efforts to make voice interoperability a thing: My colleague Kevin Tofel took a look at Amazon’s voice interoperability initiative and decided that until Google, Apple, or Samsung participated, it offers limited value to end users. This is similar to what Ecobee CEO Stuart Lombard told me in an interview last week. Ecobee is one of the 30 participants in the interoperability initiative. Kevin is also concerned that requests might be handled differently based on the assistant, leading to confusion among consumers. On the other hand, if we get a good user experience from all assistants on a hardware platform, it could give users the chance to switch between assistants without buying new hardware. That would mean that if Amazon implemented better privacy practices, a user could switch over to it as opposed to using Google. I’m sure many consumers would like that ability. (StaceyonIoT)