Buried Alive

By Francois Lamy

How to manage the influx of IoT data to improve your product.

Smart, connected products—like Amazon Echo, FitBit fitness trackers and Nest thermostats—are hot in the market right now. Given that sensors have become so cheap, more and more manufacturers are adding smart, connected capabilities to their products. These capabilities aren't just good for business, though. Data from smart, connected products in the field can be used to improve and produce better products and lead to increased market share.

However, we're already a few years into the smart, connected revolution, and it's becoming apparent that fatigue has set in. Consumers are left scratching their heads over "dumb" smart, connected products. A quick search on the internet will find you a smart hairbrush that tells you if you're brushing your hair wrong and a toaster that will alert your phone when your toast is done. With all due respect, the manufacturers that created these products are just trying to jump on the Internet of Things (IoT) bandwagon.

In addition, manufacturers that were eagerly adding numerous sensors to their products are now drowning in a flood of data. The saying "too much of a good thing" comes to mind. This influx of data is actually hindering their efforts to improve the product, and is leaving them at a complete loss as to what information is valuable and what is background noise.

Sensors, Sensors Everything
The root of the problem is our overzealousness to "sensor up." This, in itself, is not a bad thing. But data dumps can keep us from finding and analyzing the small, golden nuggets of data that will give us the insights we need. The no-brainer solution is to put fewer sensors in your product. If you're used to adding sensors to everything that moves, you might worry that cutting back will mean that a piece of key information might escape you that you had never even though of. Perhaps, but the bottom line is we need to start being more strategic about what data streams we want to get.

Unfortunately, you can't always slap on a sensor at the end of the assembly line and expect to get the data streams that you need. The more complex your product is, the more strategic you need to be with sensor placement. And to be strategic, you need to start thinking about sensor placement at the very beginning of the product-development process.

Designing for Connectivity
When you include sensors in the initial designs for your product, you'll be sure that you are capturing the correct data streams to meet your IoT ambitions. This is one case in which more is not better. Of course, this isn't anything radically new. Automotive manufacturers have traditionally added sensors in the design stage of the development process with anti-lock braking systems and the sensors that display fuel levels. The difference now is that smart, connected systems mean we can connect even more components, parts or more advanced systems. The traditional anti-lock braking system and fuel display haven't been used for other onboard functionality, or had connectivity outside of the car.

With advanced smart, connected capabilities, automotive manufacturers can use sensors in anti-lock braking systems to monitor across the fleet of a particular model to identify and fix any issues in the system before they become a public relations nightmare. They can collect fuel-consumption data from the fleet to improve fuel efficiency. Even insurance companies can get in on the action by using the data to give drivers preferred insurance rates.

When determining the data streams that you want to sensor for, it's important that all stakeholders are involved in the process. By building a pipeline of data-stream requests from throughout the enterprise, a design team can be sure that it is adding sensors that will bring the most value to an organization: whether that is product-monitoring sensors for the service team or monitoring for a subscription or "pay-per-use" model by the finance department.

Other Challenges to Consider
In addition to determining the most relevant data streams to sensor for, it's imperative that the design team also determine the best location and the exact quantity of sensors to be used. Once these are determined, the next challenge is to decide on the best type of sensor to use to capture these data streams.

Let's look at an automotive manufacturer that wants to display the outside temperature on the dashboard. The manufacturer can do this in a number of ways. It could attach a temperature sensor under the car's hood and connect it to an onboard computer, which would display the reading on the dashboard. Or it could use GPS data to determine the vehicle's location and then remotely connect to a weather service to find the location's temperature.

Ultimately, to make designing for connectivity feasible, manufacturers need to ensure that they have the proper tools: the right sensors to meet their needs; a design tool that supports the design and development of smart, connected products and systems; software to enable data flow through the different layers of the system and ensures the right information is passed through; and a way to manage all of the data from the field so that it can be used to improve future iterations of the product. You also need to be sure that you are able to relate this data to the product information that is already available, including the bill of materials (BOM) and digital product definition. To capture this information, you'll need to build a strong digital engineering foundation with a product lifecycle management (PLM) system.

With a strategic process to capture data streams, the right tools and a PLM foundation in place, your organization will be able to begin receiving and properly interpreting data from the field into your design practices. And frankly, that's just smart.

Francois Lamy is the VP of PLM solutions management at PTC. He is responsible for driving PLM solutions strategy and roadmap, and leade the company's PLM segment solution and product management team. Since joining PTC in 1998, Francois has held several product and people management positions in both MCAD and PLM organizations. He has a Bachelor of Sciences degree in mathematics and an MS degree in mechanical engineering from Arts et Métiers ParisTech, in France.