The value of IoT technology often eclipses the initial use case, and for this reason, businesses launching IoT projects need to be mindful to ensure technology can scale to accommodate data demands that may not be anticipated at the design stage, according to new research from ABI.

Two key aspects of scaling IoT projects and data processing capabilities are close attention to security and networking. That ensures, for instance, that an initial use case of collecting data from sensors to enable smart parking can also accommodate demand for leveraging those sensors to gather environmental data to assist in, for instance, emergency response.

“Almost everybody had a story of working on a POC in which they thought they were going to measure something or collect data for a certain purpose, and they ended up reinventing the whole company and it became a completely different animal,” said Kevin McDermott, principal analyst IoT at ABI Research, on talking with customers for his recent research, “IoT Data Traffic: Application and Market Analysis.”

Having spent the last two decades in Silicon Valley, McDermott wanted to look at how IoT projects could scale beyond one-off POCs, and how quickly the changes would come into effect. How IoT projects and the demands of the data analysis evolve became a central focus of his research.

In some cases examined by McDermott, those implementing IoT projects didn’t give strong consideration to security because the data involved in the specific POC wasn’t considered valuable. Such an outlook doesn’t take into account the evolving use of the data collected down the road, according to McDermott.

In turn, IoT system designers should be mindful of implementing a network provenance system framework, which allows analysis should there be a system misbehavior, according to McDermott.

By anticipating the detailed use case scenarios including data bandwidth growth, the uncertainty between design expectations and real-world experiences can be minimized, McDermott said.

The fastest moving IoT space, McDermott said, is OEM telematics. Bandwidth requirements will continue to grow here because of the demand for real-time video capability in, say, assisted and connected driving. On the flip side, he found those implementing smart city projects tend to take more time, he said, taking a couple of years to analyze the potential in the aim of rolling out a system that will run for a decade or more.

Overall, the cloud won’t be able to handle increasing bandwidth and response time requirements, and companies will increasingly look to edge computing to perform local intelligence.

“As good as the cloud is, should you lose that connectivity, you don’t want everything to stop,” McDermott said. “Having that local resource that can do everything is going to become the most important.”