Why should you learn IoT

Internet of Things (IoT)

ARI Fleet Management manages 1.2 million things with wheels between North America and Europe - from telephone company trucks to company cars to rail maintenance vehicles. The telematics sensors installed in the vehicles collect data every three to 30 seconds. "Every two weeks we get the equivalent of all the data we've collected over the past 20 years," says Bill Powell, director of enterprise architecture at the Mt. Laurel, New Jersey company.

This involves a wide range of information: ARI can use its gyroscopic sensors, for example, to detect whether the driver is making a cavalier start after a stop or is getting into the iron, the information from the engine sensors provides information on whether the machine has been idling for too long running.

One of the most fascinating and granular ways with all of these terabytes of data is to compare where a company's fuel card was used (based on the gas station's geospatial data) and where the vehicle was at that time. If the difference is more than six meters, ARI can prove that someone used the card to fill up an unauthorized vehicle.

IoT data: It all depends on the context

As the example shows, the Internet of Things (IoT) is not just about sensors and data, but also about the context in which they are used. This makes IoT a challenge for IT executives that goes beyond their area of ​​expertise and, in addition to IT, also includes operational procedures and business processes.

Longstanding IT professionals need to be forgiven if they let their skepticism run wild about the maturity of IoT. For years, they have read about sensors that help vending machines report that they are full (and therefore do not need to be refilled), or about RFID-supported supermarket shelves that help with inventory. These are interesting developments, no doubt, but not the kind of progress that has the potential to transform a business from the ground up.

This development now finally seems to be in full swing: In their outlook from May 2016, market researchers from IDC forecast that global spending on IoT will rise to 1.46 trillion dollars by 2020 - from 692.6 billion dollars in 2015. This corresponds to an average annual growth of 16.1 percent. At the same time, the number of installed IoT endpoints (ie the "things" or sensors) is expected to climb from 12.1 billion to over 30 billion in 2020.

According to Matthew Littlefield, President and Chief Analyst at LNS Research, who has conducted IoT surveys with the Manufacturing Enterprise Solutions Association for the past two years, the issue is increasingly on the radar of manufacturing companies. In 2015, 44 percent of survey participants said they were not familiar with IoT, this number fell to 19 percent in 2016.

Four lessons from IoT veterans

Computerworld spoke to a number of IoT veterans, companies from different sectors such as manufacturing, logistics, smart cities and agriculture. Almost all of them report stumbling blocks on the way, but also stated that they have achieved significant successes with their IoT investments, or at least expect them to. Here are four lessons to be prepared for.

Lesson number 1: be ready for a deluge of data

ARI Fleet Management's Powell admits that the technology his company used was immature five years ago, and in some cases limited to GPS data. “Originally it was just breadcrumbs,” he explains, “we could see the movement of a vehicle over the past five days on a map, but now, with data collected every 30 seconds, the amount is more like a jet from a fire hydrant. "

So, from Powell's perspective, data is perhaps the biggest problem CIOs struggle with. "Don't underestimate the amount of information," he warns. "Some people think that if they collect all the information available, they will have the enlightenment for business. That is not the case. Perhaps that is how you can deal with a smaller amount of information. If you try the same thing with telematics data, you will drown in it." His corresponding tip: Bring the business logic as close as possible to the sensor.

Jon Dunsdon, CTO at GE Aviation, agrees with Powell. His company has been monitoring jet engine data for 20 years. "One of the challenges is that only a few planes are sending data continuously. You do the initial analysis on board and then after landing the data is sent," he explains. Ten years ago that was 3.2 kilobytes of information, today hundreds of megabytes are received when the plane lands.

At the same time that GE Aviation is collecting information from the aircraft, the company is also collecting other information. The aim is to offer customers - including United Airlines and Southwest Airlines - added value. "We look at things like planning data, weather information, no-fly times, etc.," says Dunsdone. "What impact does it have on planning if there is a storm in Chicago? How can costs and the extent of disruption be reduced?" Dunsdon uses a data lake to house the amount of analysis and increase performance, and recommends other IT executives to do the same.

Chemical giant BASF has two billion-euro production facilities in Ludwigshafen, known as steam crackers, the size of several soccer fields. In the crackers, raw gasoline is broken down into shorter molecules such as ethylene or propylene, which are the basic building blocks for further production. The systems are usually in operation around the clock and the group uses sensors to calculate the downtimes, which cost several million euros per hour.

According to Wiebe van der Horst, Senior Vice President Process & Enterprise Architecture at BASF (and CIO of 2014), BASF was warned of a possible failure at best a few hours in advance before using these sensors, but this has now changed. "We worked on artificial models with the help of various algorithms in order to establish a correlation between the activity of the machine and its maintenance data," explains van der Horst. After analyzing the data with SAP Hana, BASF can identify potential downtimes a few weeks in advance. "These are valuable insights for us," said the BASF manager. "We take the results and apply them to other factories around the world."

Without the ability to combine unstructured and structured data, the project would not have been possible. And that's not all: "Not so long ago it would not have been economically viable to carry out this type of calculation and analysis," explains van der Horst.