Subscribe / Unsubscribe Enewsletters | Login | Register

Pencil Banner

Predictive data, the real workhorse behind the Internet of Things

Nancy Gohring | June 12, 2014
Companies are having to be pickier about which data sets they collect, process and use. It's all about having the right architecture in place.

Putting the pieces together

"We see countless companies that are past the part of experimentation and deploying sensors and collecting data" but that don't have a fully integrated solution, IDC's Turner says. "It's the complexity of the implementation."

Businesses need infrastructure on the back end that enables the combination of data from various sources as well as the analytics power to make sense of it all. Then they need dashboards or visualizations that let line of business people understand the meaning of the data so they can make smart decisions based on it, he says.

Daikin Applied is one company that, with the help of partners, has put together a sophisticated set of hardware and software that collects and analyzes 4,000 different data points about its commercial heating and air conditioning rooftop units. The system, designed with Intel, syncs with weather forecasts to allow building owners to adjust for changing temperatures in advance and lets Daikin know when changes in energy use by individual components indicate a failure is imminent so that the company can dispatch a repair technician beforehand.

In the future, the system also will let Daikin feed important data to local utilities that might be able to use it to reduce the power output to any given piece of gear. Talks with utilities are in preliminary stages right now, says Kevin Facinelli, executive vice president of operations at Daikin Applied. (Daikin Applied is part of Daikin Industries, the largest HVAC manufacturer in the world.)

In this implementation, hardware plays an important role. The system starts with a gateway that's based on Intel's Quark system on a chip (SoC), runs Wind River's operating system and is secured by McAfee software.

"Instead of just passing all the data through to the cloud, we have an SoC so we can do pre-possessing," Facinelli explains. That means the gateway, which will be built into all future Daikin rooftop systems, sends only important data, like a change in status of a component, rather than sending along an endless stream of "I'm normal" signals, he says. Doing some processing on site reduces the volume of data that needs to be transmitted -- Daikin primarily uses cellular connectivity -- and also helps to reduce the data warehousing load on the back end.

Daikin also uses a power meter that monitors the supply coming into the unit. Via the gateway, it sends data about the power signal to an Intel cloud, where it's analyzed to determine the power usage of each component inside the HVAC system, like fans and refrigerant compressors.

Without the back-end analytics, Daikin would have to install meters on each component, an implementation that would be prohibitively expensive, Facinelli says.


Previous Page  1  2  3  4  5  Next Page 

Sign up for Computerworld eNewsletters.