|
Did you receive this newsletter as a forward? Subscribe here
|
|
| The IoT needs a huge helping of trust if it wants to work |
| |
| By Stacey Higginbotham |
I don't know about y'all, but I'm hitting that point in the summer where I just want to hang out, read books, and drink a cold beverage while watching the sunset. Luckily, the days are long, so I get to spend time relaxing (at least until the wind shifts and smoke from the nearby wildfires hits) and thinking about the future of the IoT.
This week, I've been thinking about trust. That's in the wake of reading one story about the police in Chicago changing data in the ShotSpotter gunshot detection system and another about a researcher who was struggling with the Apple Watch algorithms and how they might affect his work. Both stories are centered around trust, a crucial element that gets largely lost in the IoT— specifically trust in the data and trust in the algorithms. |
|
| — During last summer's protests the ShotSpotter system had to distinguish between guns, fireworks, and more. |
|
|
We talk a lot about security and privacy in the IoT, but very little about how data is gathered, who has access to it, as well as how it's authenticated and kept from being compromised. Meanwhile, when it comes to algorithms we don't always know where the training data came from, how the algorithms were built, or how they change in response to real-world experience.
The answers to these questions have huge implications because the data and algorithms are driving public policy. They can also be used to determine credit scores, inflate the price of goods — even determine who gets health care. Not to mention that some data can drive machines to action, such as in the case of an irrigation system or manufacturing plant.
It's why we need to develop methods for data assurance and attestation so we understand how a sensor or device generates data. We also need to build chains of custody for data as it moves throughout computing systems. And then we need to figure out how to build algorithms in ways that are both replicable and transparent.
What might this look like? First, let's consider two of the most recent examples of a trust failure related to data. With the ShotSpotter system, court filings allege that the Chicago Police Department changed the classification of certain sounds in the system from a firework to a gunshot. Later, another analyst working for ShotSpotter changed some of the location information to corroborate a story that the police department told with regard to a police shooting of a 13-year-old boy. From the Vice article:
But after the 11:46 p.m. alert came in, a ShotSpotter analyst manually overrode the algorithms and “reclassified” the sound as a gunshot. Then, months later and after “post-processing,” another ShotSpotter analyst changed the alert’s coordinates to a location on South Stony Island Drive near where Williams’ car was seen on camera.
Through this human-involved method, the ShotSpotter output in this case was dramatically transformed from data that did not support criminal charges of any kind to data that now forms the centerpiece of the prosecution’s murder case against Mr. Williams,” the public defender wrote in the motion.
The prosecutors working for this state decided to withdraw the ShotSpotter evidence rather than explain the changes.
In the second example, JP Onnela, associate professor of biostatistics at the Harvard T.H. Chan School of Public Health, wanted to use data from the Apple Watch showing heart rate variability. He chose to use data from a time period between early December 2018 and September 2020. He pulled the data twice, seven months apart. However, when comparing the results from the HRV data pulled earlier he discovered they were statistically different from the HRV data that he pulled later. In other words, the way the Apple Watch calculated heart rate viability had changed — and ultimately had changed enough to make Onnela question his use of the Apple Watch for his research.
Most doctors and researchers are aware of the foibles of consumer wearable devices, namely their lack of accuracy and their changing algorithms. But as Apple, Google, and Amazon continue pushing these devices for wellness and even on-the-job monitoring, it's worth understanding how those algorithms change and who those changes might advantage.
When it comes to the data itself, we need ways to ensure that a sensor is calibrated correctly and that it isn't compromised. The National Institute of Standards and Technology (NIST) and several other standards bodies exist to ensure that sensors meet the required specifications, but not all sensors meet those standards. Second, the IoT system needs a way to ensure that the data it ingests come from an authorized sensor that is telling the truth.
Then we need chains of custody that ensure the sensor data isn't changed within a system. And finally, we need ways to audit the algorithm processing the data to ensure that it is fair and that it meets the public objective. For example, in the Shot Spotter example, perhaps the need to keep a clear chain of custody would have prevented analysts from reclassifying data or would have forced ShotSpotter to clarify why that happened.
ShotSpotter has denied that its analysts changed evidence to fit a police narrative and instead noted that it always has analysts create a separate record for court proceedings and that Vice, in its story, conflated the two separate events. However, the statement from ShotSpotter doesn't address Vice's characterization of the original misclassification, that of turning the fireworks into a gunshot.
The point is that right now, we like to think of data as holding some eternal truth, when in fact it is as biased as the people trying to use it to set policy, monitor gunshots, or promote our health. Without mechanisms to establish trust in the sensors, the data, the way that data gets turned into an insight, and the algorithms themselves, objective, data-driven decisions are as much a chimera as objective journalism.
|
|
|
|
|
| Your guide to leading successful IoT initiatives SPONSORED |
| |
|
Successful IoT projects can transform every aspect of your business. In this guide, discover the processes you need to follow to make your IoT product vision a reality. You’ll find:
- Best practices and frameworks for leading IoT initiatives
- 10 key steps to an Agile IoT release
- What it’s like to work with an IoT development firm
Download now
|
|
|
|
|
| Elipsa and the promise of a no-code future |
| |
When it comes to digital transformations — or even simply trying to use sensor data to optimize a business process — there aren't enough data scientists to go around. And even if there were, some problems probably aren't worth the time and cost of getting an expert statistician involved.
This is where a bevy of startups offering no-code solutions come into play. One of these, Elipsa, was formed two years ago with an eye toward providing data analysis for financial data, but by the end of last year had switched to providing analysis for IoT data. |
|
| — The Elipsa software runs sensor data through multiple neural networks to figure out which equation works best for your data. Image courtesy of Elipsa. |
|
|
Elipsa makes software that takes in data and runs it against several statistical algorithms and neural networks to figure out what math is most effective at finding anomalies or trends in said data. To use the software, an employee at an Elipsa customer provides the data and then tells the software what they are looking for, such as certain anomalies. Or perhaps they want to forecast inventory availability, or pricing trends.
After sharing the data and defining the types of results, Elipsa's software runs through several different algorithms to figure out which ones give the most accurate prediction. The software shows how confident the chosen algorithm is in the prediction and which pieces of data are most important in reaching the prediction.
During the demo, I was impressed at how easy it seemed to be to use and how well it communicated both the prediction and where the predictions might break down. Each prediction gets turned into an API that the customer can then use to export the models and predictions to other software or services.
By making it so easy, individual machines could potentially get their own individual APIs and tailored predictions as opposed to having a manufacturer try to apply one algorithm to a bunch of different machines, each running under very different circumstances. Elipsa charges per model, which means that every additional machine or API feed generates revenue.
Jeff Kimmel, a co-founder and the CEO of Elipsa, said that demand skyrocketed during the pandemic because so many companies tried to add remote capabilities and sensor-based tracking systems for employees. Plus, companies in the industrial sector or asset tracking are underserved by data scientists when compared to the finance industry, which also drove the pivot.
Elipsa is a small company with only five employees and plans to raise funds. However, it's already inked partnerships to provide its software along with Losant, Software AG, and Crosser. Brandon Cannaday, the chief product officer at Losant and one of its co-founders, said that at least one customer is testing the Elipsa product in its operations.
Losant provides equipment and limited services for customers building out edge computing. Cannaday said that most customers have some type of AI aspirations, but they don't have the data scientists on hand. In many cases, they may not even yet know what they want. With Elipsa, they can start down the path of getting insights in an easy and replicable way without spending a lot of money on data scientists or consultants. For many customers, this is enough, especially as they start their digital transformation efforts.
It's worth noting that what Elipsa does wouldn't be possible without first establishing trust in both the data and the methodologies for creating algorithms. And once companies build trust in the IoT, others can then abstract some of the hard work of turning data into insights. If a business can rely on the data and the math that drives a decision, then it's much easier to hand that over to experts and let executives or line operators keep doing what they do best. |
|
|
|
|
| Build the future with Very SPONSORED |
| |
|
Develop new smart technologies or expand the capabilities of your existing products with a battle-tested and trusted IoT partner. Talk to our team today to discover how Very can amplify your current product offerings.
Get started. |
|
|
|
|
| Episode 331: Safe words for smart homes and cheap mesh |
| |
We start this week’s show with a $200 million funding for Wiliot, a company I profiled back in 2017 as one of the vanguards of low-power sensing. Then we tackle a creative idea that could see consumers create safe words for their smart homes to indicate when they might be in trouble. Next up is President Biden’s National Security Memorandum on securing cyberinfrastructure. Like coffee? This connected coffee machine raised $20 million. If coffee’s not of interest, perhaps you’ll want to hear about research into the incidental users of smart home gear and what we owe them, or how to change Alexa to Ziggy and get a new voice option. I also talk about a new dev kit that will let you hook up Swarm’s satellite connectivity to a variety of sensors. Or maybe you’d like to hear Kevin’s review of the $59 Vilo mesh Wi-Fi system or about the upcoming Firewalla Purple device.
|
|
| —The Swarm Eval kit could be yours for $499 plus the $60 annual connectivity fee. Image courtesy of Swarm. |
|
|
Our guest this week is Jason Shepherd, the VP of Ecosystem with Zededa, a container orchestration company for the industrial internet of things. It’s been a while since Shepherd has been on the show, so I asked him for an update on the IT and OT divide that we talked about four years ago. Both sides are coming together, but there are still challenges when it comes to bringing IT to scale in operations. We talk about heterogeneity, security, the challenges of remote access, and more differences worth thinking about when we put computers in industrial equipment. We also talk about the challenges of scaling machine learning models at the edge, and especially those designed to adapt to changing real-world conditions. It’s a fun interview.
|
|
|
This week on the IoT Podcast Hotline, we answer a listener question about the Firewalla Purple.
The IoT Podcast Hotline is brought to you by Silicon Labs' Works With event. Works With 2021 is your gateway to the latest IoT trends, training, demonstrations, and workshops from the biggest names and ecosystems in the industry. September 14-15 online. Register for free. |
|
|
|
|
|
| Will this $59 mesh network system work for you? |
| |
|
| — Kevin tested the $59 Vilo mesh Wi-Fi system and found it was a good choice for people who weren't focused on extreme speeds or latency. If you just need better coverage and can live without all the bells and whistles, this is a good option. Image courtesy of K. Tofel. |
|
|
|
|
|
|
|
| News of the Week |
| |
Silicon Labs has completed its divestiture of the automotive division: In April, chipmaker Silicon Labs said it would sell its infrastructure and automotive chip division to Skyworks for $2.75 billion in cash. Silicon Labs initiated the transaction so it could focus solely on providing radio and microprocessors for the IoT. The company's IoT business comprised 58% of last year's revenue while the infrastructure and automotive business comprised 42%, but the growth in the IoT business represents a greater opportunity, according to CEO Tyson Tuttle. Also this week, Tuttle said he plans to retire in January 2022, and Matt Johnson, the current president, will step up and take on the role starting that year. To hear Johnson on the Skyworks deal, check out this podcast from April. (Austin American-Statesman)
Batteryless IoT sensor company Everactive raises more money: This is a big week for low-power chip startups. First Wiliot scored $200 million, and now Everactive has raised an additional $16 million from 3M, Ericsson, and Armstrong International, a thermal utilities company. Like Wiliot, Everactive has turned its chip-based low-power sensing tech into a full-on service, providing the software, the sensor, and the insights for customers that want to monitor their equipment without changing out batteries all the time. I love these startups because without them, we're stuck with a much more limited IoT, one constrained by wires or constant battery changes. (Everactive)
Smart sensing for better building ventilation: Air quality monitoring company Airthings has signed an agreement with Edwards, a division of Carrier, to provide indoor air quality monitoring for commercial buildings. Edwards will resell Airthings' business software service along with its fire and safety products. The Airthings for Business product lets building facilities operators monitor, visualize, and control indoor air quality. The deal also sends Airthings data into Carrier's Abound in-building data platform. With businesses focused more on proper ventilation in the wake of the pandemic and another summer of wildfires, solutions like Airthings are going to proliferate. In Europe and in Asia, indoor air quality has been a much more active market, but we'll see more attention to it here in the U.S. as well. (Airthings)
Smart underwear gets Health Canada's approval: Underwear that monitors patient ECG data has been approved by Health Canada and is in line for FDA approval in the U.S. While the undergarments from Myant's Skiin brand are approved for tracking ECG data, they can also provide other information such as heart rate, heart rate variability, and core body temperature. Myant also plans to add sleep tracking and location tracking to the garments. The sensors are embedded in the fabric of the bra and T-shirt and connect with a removable puck for the electronics and radio. (Wearable Technologies)
Siemens and Dow create a chemicals manufacturing testbed for digital twins: Process manufacturing is a huge industry that covers everything from oil refining to making the filling for Twinkies. And for years the sector has been highly automated, with companies such as Honeywell and Emerson providing connected sensors and controllers to manage the various processes involved. As the focus on IoT has increased, process manufacturing companies are now trying to use data not only to make new chemicals or products but to monitor the machines as they make those chemicals or products. To make manufacturing more efficient, companies are trying to build digital replicas of the manufacturing process and the machines governing that process. The partnership between Dow and Siemens is designed to show that digital twins can be created, but also that they can provide value that justifies the effort of building them in the first place. (Venture Beat)
Will attention-based neural networks enable computers to see as humans do? This is a fascinating piece because it's a reminder that computers ultimately process information differently from the way we do, and those differences are usually obfuscated by clever marketing teams and a lot of computing power. But this story argues that teaching computers to change the way they analyze an image could require less computing power and result in a computer vision algorithm that uses less power. I love this stuff! (EETimes)
Where is Thread on Eero's routers? A few weeks ago, Kevin enabled the Thread radio in his Eero Wi-Fi 6 routers, excited to see how it would affect the Thread-capable devices in his home. But when viewing the Thread network diagrams (the Eve app provides a great one!), nothing actually changed. We now have an explanation from a product manager at Eero as to why, and who also says that Eero will start announcing specific products that will use Eero's Thread radio in the coming weeks. He adds that if you're building a connected Thread device and want it to work on Eero to reach out to him. (Reddit)
Hear about how hackers are going after industrial controls systems: If podcasts are your jam, check out this interview with Armis CISO Curtis Simpson, who talks about how hackers are devoting more attention to hacking industrial operations. While once a focus of nation state-style attacks, the rise of ransomware and potential for profit is attracting others to the field. (Threatpost)
We'd appreciate your support. To sponsor this newsletter, please request a media kit. |
|
|
|
|
|
|
|
|
|
