Internet Of Things: The Rubber Meets The Road

Editor's note: John Morris, Principal at Business Decision Models Inc., co-wrote this article with Chris Taylor.

The Internet of Things (IoT) is here. We've moved beyond speculation about what IoT might look like to the place where the rubber meets the road: that is, to where IoT demonstrates real business value.

This evolution raises questions for how to best manage and benefit from the technology as it continues to grow in ways that we may not realize.

An IoT Business Case

There's a clear business case for efficiently managing 73,000 refrigeration units. Using facilities and equipment management software platform from Verisae, Inc., a major European retailer has deployed a predictive analytics-based program for refrigeration cabinet maintenance, covering over 1,000 of its retail locations.

As data streams in from sensors on the units, predictive analytics decide what’s likely to break in the near future. Refrigeration technicians are dispatched with everything required for a non-disruptive, preventative maintenance call during the two-week window prior to a predicted failure. According to the company, the system runs with up to 98% accuracy.

As a result, its key performance indicators are encouraging; refrigeration maintenance costs are down, equipment "trading time" is up, business risk is down, offered product quality is up, and, of note, end consumer satisfaction is up for the retailer.

Three IoT Challenges

While the Verisae case is a clear win, IoT and the ever-growing abundance of data still present significant challenges.

These include the complexities involved in automated decision-making, the need to conceive entirely new use cases, and the difficulty of finding the right data at the right time.

Let’s look at each challenge in more detail.

Automating decisions: The engineering and analysis work for automating decisions requires significant investment, and will likely take multiple years to execute in any substantial domain, especially if starting from scratch.

The refrigeration project mentioned above was three years of work. The project team involved three parties; software and services vendor Verisae, a systems integrator specializing in facilities management, and retailer IT and business staff, all working together. 

Among its tasks, the team needed to construct predictive algorithms for refrigeration systems maintenance, incorporating data from multiple telemetry sources and sensors.

A critical requirement was to avoid false and duplicate alarms, because a field service call based on a false positive is expensive. This presented a significant challenge in bringing the project to life.

Automating decisions has advantages and risks, in part because computing currently lacks the same discernment capabilities of the human brain.

Limits of imagination: When automating complex tasks, we're usually working within familiar paradigms. As we define more and more of our world in digital terms pulled from ubiquitous sensors, the most familiar use cases will be replaced by completely new ways of doing business.

Reimagining our world is difficult, and the gap between computing and the human brain remains a significant barrier in thinking of new ways to use these technologies.

Dr. Carmen Simon is a cognitive neuroscientist looking at this very problem. She noted the following points:

"In order for machine intelligence to be fully formed, mimicking human intelligence, it needs the ability to sense the environment, which it can do now; become self-aware, which machines are getting better at doing; and predict the future, meaning setting goals and plotting strategy."

Dr. Simon also noted, "We may be almost there, but this is where the machine needs more help by learning from the human brain, which is constantly on fast-forward."

"There is little adaptive advantage for the brain to be in the moment, meaning our brains can’t help but look forward. If it seeks intelligence, the machine must catch up."

Fast data: As more data is created and moves faster, finding and reacting to the right data becomes harder. Software can solve a significant part of this problem using techniques such as filtering and applying logical rules to data before it ever reaches storage, allowing data to “stream” through systems while being carefully monitored.

TIBCO's Mark Palmer noted "Today's increasingly digital businesses require the ability to work comfortably with 'Fast Data' to survive."

These three areas represent just part of the challenge facing IoT innovation driven by the arrival of the data age.

Want to engage in a deeper discussion on IoT? Please join me and John Morris for the Internet of Things Summit at Interop Las Vegas on April 28. You can hear directly from Dr. Carmen Simon and TIBCO's Mark Palmer on the opportunities and challenges of IoT. We hope to see you there.