The definition of the Internet of Things in simple terms is: A network of devices that contain embedded technology to communicate, sense, and interact with other devices.

In short, the Internet of Things is a way for everyday objects to talk to each other and to talk to you, such as when a refrigerator alerts a mobile phone that the milk has run out. There are already plenty of impactful use cases around the Internet of Things.

Let us be absolutely clear: the Internet of Things and big data are not the same thing, whatever media and marketers might claim. By way of clarification, consider this: a European engineering firm designed and implemented an Internet of Things use case for a European railway operator.

Using a low-bandwidth channel, the locomotives running on diesel would regularly send just one number representing the diesel level in the tank. Even assuming a few hundred trains operating on the network, this does not amount to more data than 10 kilobytes of data per hour.

It is a beautiful example of a small data Internet of Things with very simple analytics but very high ROI, as the owners of the trains can collect the information and use that in negotiations with the operators, who are leasing the trains from them.

The data helps the owners understand how often the trains are idle or out of service.

Even the humble train can benefit hugely from IoT technology

Most of the data going between devices comes in the form of simple sensor readings, such as: locations; temperatures; levels of some liquids; indications of the presence or absence of some object; mechanical factors such as forces, speeds, accelerations, tensions, vibrations, and pressures; and electrical factors such as currents, voltages, magnetic fields, and light.

Such simple data still needs a lot of processing to make sense of it. Just collecting it does not help.

Higher-level data is already preprocessed locally inside the device. For example, an electricity meter could send meter readings, the state of health of some machines, or some simple diagnostics analyzed by software. Noncentral processing has the advantage of the raw data getting compressed into a useful finding that another machine can use.

The drawback is that local electronics become more expensive. For example, a sensor in a tire is pretty dumb, as it cannot cost too much and the physical forces inside a tire are very demanding for normal electronics, but a smart meter might already contain some significant amount of local processing power.

If you would like to get some hands-on experience with the world of Internet of Things, buy a Lego Mindstorm and program some simple use cases, such as making the Mindstorm robot follow a carpet line in your living room. You will quickly understand the programming of the sensors and actors. Lego Mindstorm can teach you the fundamentals of the Internet of Things in less than a week.

The forecasted numbers of devices are staggering. The GSMA estimates 24 billion connected devices by 2020, half of which will be mobile.

Cisco, the largest networking provider in the world, estimates the number of connected devices to reach even 50 billion by 2020, implying that there will be more than six connected devices per human.

If we further know that the combined population of Organization for Economic Cooperation and Development (OECD) countries is only about 15 to 18 percent of the global population and that most connected devices are going to be in the OECD countries in the medium term, the average number of connected devices per OECD inhabitant will be more like 30 to 40.

People will soon have more connected devices than they can reasonably carry about

Just think of your own home: refrigerator and deep freezer, phones, tablets, laptops, music and entertainment systems, climate control, garage, car and garage doors, lights and shutters, mailbox, security systems, pet feeders, and drones… I am sure I missed a few.

Any fully connected home can easily have more than 100 sensors and connected devices. Do we need all of this? That is probably more of a philosophical question! But this is the world we have.

What are the application areas for the Internet of Things? As much as I feel that marketers overuse the word smart—since many of the devices don’t actually learn or do more than basic data cleansing—it does apply to the potential of this technological concept in four areas:

1. Smart life: Making our lives simpler and safer. This includes services in our homes, personal healthcare, food and nonfood retailing, banking, insurance, and private and public services.

2. Smart mobility: Making transport faster, more enjoyable, and more reliable. This includes the connected car, urban and multicity traffic management, payment solutions, distribution and logistics, and fleet management of trucks, cars, trains, airplanes, drones, and so on.

3. Smart cities: Managing city infrastructure more efficiently by connecting the inhabitants to make them safer, running grid management and smart metering to provide more efficient utilities, and improving facilities management and waste management to make cities cleaner.

4. Smart factories and supply chains: Improving manufacturing and supply chain processes to lower cost and improve uptime and quality. This includes preventive maintenance, better process control and compliance, better planning through decision support, faster time to market for manufactured devices, and better integration of whole supply chains and market demands. Industry 4.0 promises depend a lot on the Internet of Things.

This is an edited extract from Mind+Machine: A Decision Model for Optimizing and Implementing Analytics by Marc Vollenweider (Wiley, December 2016).