Internet of Things · IoT
Internet of Things
What is the Internet of Things?
- Connecting things to the internet
- Exchanging data and remote control
- The foundation of smart homes and factories
What IoT really changes is this: things you once had to go and see in person, the data now tells you first. In the past, to know the state of a machine, someone had to walk to the site, read the gauge, and write it in a log, and that log was only reopened after an accident. Once sensors and communication modules are attached, status values accumulate by the second, and decisions rest on records instead of individual intuition. That is why success depends not on how many sensors you install, but on what you will decide differently once you see the data.
A deployment usually has four layers: sensors and actuators, gateways that collect and forward data, the communication network, and the storage and analytics platform. The right connectivity depends on the use case. Wi-Fi and BLE suit short distances inside a factory, while LTE-M, NB-IoT, and LoRaWAN suit large sites and outdoor assets. Assets that must run for years on a battery call for longer transmission intervals. Data that arrives thousands of times per second, such as vibration, is not all sent to the cloud; edge processing, a first pass at the gateway, is effectively standard.
It also helps to know how IoT differs from similar terms. M2M is closed communication in which machines exchange data over dedicated lines, while IoT is an architecture in which multiple systems connect over internet standards and open platforms. Layer real-time data onto a 3D model and run simulations, and you get a digital twin. IoT is the layer that produces the raw material for the technologies above it, so if data formats and tag names differ from one machine to the next, analytics and prediction will be just as unreliable.
The most common misstep in the field is “let’s just install sensors first.” The more data points you collect, the more you pay each month for connectivity and storage. If no decision changes based on those values, no one will open the dashboard a year later. Security is the same story. A single gateway left on the factory-default admin/1234, or a single control network connected to the office network, becomes an entry point that can shut down an entire line. It is safer to start by narrowing the scope to a few machines and a few data points, covering only what can deliver a numerical answer within six months.
Ilyoung JungExpert in smart factories, equipment management, and Industry 4.0
John KimInstructor John Kim, author of “B2B Marketing Design” and CEO of Nanum ConsultingIn 1999, Britain's Kevin Ashton coined the term "Internet of Things" while presenting an idea at P&G to track inventory automatically with RFID.
Consider a hypothetical case. Company A is an auto-parts injection molding manufacturer with 320 employees and annual revenue of KRW 118 billion (≈ US$88,283,700). Among its 48 injection molding machines, those over 10 years old repeatedly showed mold temperature anomalies and sudden stoppages. Last year, unplanned downtime totaled 74 hours, with losses of about KRW 300 million (≈ US$224,500). Executives launched an IoT project with the goal of “cutting downtime in half within six months.”
- Mr. Kim, the maintenance team lead, didn't instrument all 48 machines. He picked only the 12 with the most downtime history and fitted three sensor types: vibration, current, and mold temperature. The cost was about KRW 1.8 million (≈ US$1,350) per machine.
- The head of production insisted that “if 100 alerts come in a day, eventually nobody looks at them.” So for the first three months, the team set no thresholds. They simply collected data and traced back the patterns from the days when stoppages occurred.
- Ms. Lee, an IT manager, placed the gateways on a separate VLAN isolated from the office network and replaced every factory-default password. She also narrowed vendor remote support to a single VPN path.
- Using the three months of data, they kept just one rule: “if bearing vibration RMS stays at 1.8 times normal for 30 minutes or more, request maintenance.” Alerts dropped to an average of about 3 per day.
- Seven months in, unplanned downtime on the 12 target machines fell from 74 hours to 21 hours on an annualized basis, and the defect rate from mold temperature deviation dropped from 2.4% to 0.9%. Company A decided to extend the system to the remaining 36 machines.