Machines that report on themselves
The Internet of Things (IoT) refers to everyday objects that have sensors, a small processor and a network connection, so that they can send information or receive instructions. The idea took shape slowly. Long before the name existed, engineers connected equipment to networks so it could be monitored from a distance, and a few playful projects among computer scientists showed how little was needed to put an object online.
The campus vending machine
One of the best-known early examples was a Coca-Cola machine in the computer science department at Carnegie Mellon University, in the early 1980s. Graduate students were tired of walking to the machine only to find it empty or to get a warm bottle. They wired the machine’s front lights to a computer in the building, which could be reached over the university’s network. A person could then check which columns were stocked and how long a bottle had been inside, to judge if it was cold. The system is often described as the first internet-connected appliance, though the precise claim depends on what counts as the internet, as it was initially a local network that was later linked outward. Other university groups built similar systems.
Toasters and coffee pots
At a trade show in 1990, John Romkey and Simon Hackett showed a toaster that could be switched on and off over the internet using TCP/IP. It was a stunt, but it made a point: ordinary appliances could be addressed like computers. Another widely cited example came from the University of Cambridge computer laboratory in 1991, where researchers pointed a camera at a coffee pot in the Trojan Room so people elsewhere in the building could see whether the pot was full. The image was later made available on the web in 1993, and many people viewed it from around the world. It is often described as one of the first webcams.
Ubiquitous computing
In 1991, Mark Weiser of Xerox PARC wrote an influential article about what he called ubiquitous computing, imagining computers so small and plentiful that they fade into the background of daily life. His group built prototype tabs, pads and boards that connected wirelessly. The idea gave researchers a vocabulary for devices in the environment rather than on a desk.
Tagging objects: RFID and the name
A parallel stream of work came from tracking goods. Radio-frequency identification (RFID) tags contain a small chip and antenna and can be read without a direct line of sight, unlike the bar patterns read by barcode scanners. RFID was used for identifying animals, controlling building access and tracking railcars, and it became cheaper in the 1990s.
Kevin Ashton, working in brand management at Procter and Gamble, began using the phrase Internet of Things in 1999 in a presentation about linking RFID tags to the internet. He argued that computers depended on people to enter data, and that connecting objects to the network would let them report their own location and condition. Later that year, he helped to found the Auto-ID Center at the Massachusetts Institute of Technology, which worked on standards for tagged goods and a shared network that would let readers look up information about a tagged item. Ashton has said the phrase was meant partly to catch the attention of company executives; the broader idea had earlier forerunners, and no one person produced it.
Sensor networks
In the same years, university researchers worked on wireless sensor networks: many small, battery-powered devices that gather readings and relay them across the group to a base station. Projects at Berkeley and elsewhere in the late 1990s explored such devices for monitoring habitats and buildings, and a project called Smart Dust imagined sensors the size of a grain of sand. Short-range radio standards such as Bluetooth, introduced around 1998 to 1999, and IEEE 802.15.4, which underlies Zigbee, gave low-power devices a way to communicate.
What had to change
Several things held back large-scale connected objects before the 2000s. Processors, radios and sensors were costly and used a lot of power. The internet’s standard address space was limited, and attaching billions of devices would need many more addresses. Wi-Fi and cheap mobile data had not yet spread widely. As the cost of electronics fell, and as silicon chips gained wireless radios built into them, the conditions changed. By around 2010, industry writers were using the phrase widely, and forecasts of tens of billions of devices were common, though such predictions have varied greatly and are better treated as estimates than facts.
Measuring the early period
Looking back, the first decades of the Internet of Things were mostly experiments and specialist systems: industrial monitoring, logistics, utility meters and laboratory prototypes. What the early examples show is the central idea that has persisted ever since, which is that a physical object can carry a small amount of information about itself to the network and that people or software can act on it.