The Internet of Things (IoT) is the extension of the Internet to physical objects that sense, act and communicate, such as meters, thermostats, industrial machines, vehicles and medical devices. Each device carries a sensor or actuator, a small processor and a network interface, and it exchanges data with other devices or with cloud services through standard Internet protocols, usually with little or no human interaction.
Within Internet fundamentals, IoT matters because it changes who the typical endpoint is. Instead of a person at a browser using the World Wide Web, the endpoint is often a battery-powered device with a few kilobytes of memory that sends small messages at intervals. This shift places new demands on addressing, bandwidth, security and device management across the wider network.
How the Internet of Things is built
An IoT system is a layered architecture that moves data from the physical world to software that interprets it and, often, back again as a command. Engineers usually describe four layers: the device, the connectivity link, an edge or gateway tier, and the cloud or data center platform.
The device layer measures temperature, vibration, location or power use and may switch relays or motors. The gateway translates short-range radio protocols into IP traffic, filters data and buffers it when the uplink fails. The platform stores readings, runs analytics and pushes firmware updates and configuration to the fleet.
The table below summarizes the role of each layer and the technologies typically found in it.
| Layer | Main role | Typical technologies |
|---|---|---|
| Device | Sense and actuate | Microcontrollers, MEMS sensors, relays |
| Connectivity | Carry data from the device | Wi-Fi, Bluetooth Low Energy, LoRaWAN, NB-IoT |
| Edge or gateway | Translate, filter and buffer | Industrial PCs, routers with local compute |
| Cloud platform | Store, analyze and manage | Message brokers, time-series databases, device registries |
IoT connectivity and protocols
IoT connectivity is chosen by balancing range, data rate and energy consumption, because no single radio technology optimizes all three. A sensor that must run for five to ten years on one battery cannot use the same link as a security camera streaming video.
The following comparison shows the main link options and where each fits best.
| Technology | Typical range | Data rate | Power profile |
|---|---|---|---|
| Wi-Fi | Up to about 50 m indoors | Tens to hundreds of Mbit/s | High |
| Bluetooth Low Energy | About 10 to 100 m | Up to 2 Mbit/s | Very low |
| Zigbee and Thread | About 10 to 100 m per hop, mesh | 250 kbit/s | Very low |
| LoRaWAN | Several km, up to about 15 km rural | About 0.3 to 50 kbit/s | Very low |
| NB-IoT and LTE-M | Cellular coverage | Tens of kbit/s to about 1 Mbit/s | Low |
Above the link layer, devices use lightweight application protocols. MQTT is a publish-subscribe protocol over TCP in which devices send messages to a broker, which forwards them to subscribers. CoAP is a compact REST-style protocol over UDP designed for constrained nodes. 6LoWPAN compresses IPv6 headers so that low-power radios can carry IP packets, which supports the principle of end-to-end IP addressing.
Addressing is a structural issue for the Internet of Things. Billions of devices exceed what IPv4 address space can serve directly, so many deployments sit behind NAT, while IPv6 provides enough addresses for every device to be globally reachable if policy allows it.
How IoT differs from related concepts
IoT overlaps with several older and newer ideas, and the distinctions are mostly about scope, connectivity and purpose. Treating them as synonyms leads to poor architecture decisions, for example choosing a closed industrial bus where interoperability with cloud services is required.
| Concept | Core focus | Key difference from IoT |
|---|---|---|
| Machine-to-machine (M2M) | Direct device communication | Often point-to-point and proprietary, not Internet-wide |
| Industrial IoT (IIoT) | Factories, utilities, logistics | A subset of IoT with strict reliability and safety needs |
| Cyber-physical systems | Tight control loops | Emphasizes real-time control rather than networking |
| Edge computing | Processing near the data source | A deployment model that IoT systems often use |
| SCADA | Supervisory control of plants | Legacy architecture now frequently connected to IoT platforms |
The distinguishing feature of IoT is the use of shared Internet infrastructure and protocols. Device traffic travels through an ISP, across the Internet backbone and through peering or IP transit links, just like any other packet-switched traffic.
Designing and operating IoT deployments
Successful IoT deployments are defined less by the device than by lifecycle management: provisioning, monitoring, updating and eventually retiring thousands of units in the field. A device that cannot be updated securely becomes a long-term liability once a vulnerability is found.
The decision table below links common requirements to a sensible design choice.
| Requirement | Recommended choice | Reason |
|---|---|---|
| Multi-year battery life | LoRaWAN or NB-IoT with sleep modes | Radio stays off most of the time |
| Low-latency local control | Edge processing on a gateway | Avoids cloud round trips |
| Large fleets in the field | Over-the-air updates with signed images | Patches without site visits |
| Intermittent connectivity | Store-and-forward buffering | Prevents data loss during outages |
| Multivendor smart home | Matter over Thread or Wi-Fi | Common application layer across brands |
Operators also plan for data volume and cost. Ten thousand sensors reporting every minute generate over 14 million messages per day, so payload size, reporting interval and local aggregation directly affect bandwidth bills and storage requirements.
Security risks and common mistakes
Security is the most persistent weakness of the Internet of Things, because devices are cheap, numerous, physically exposed and often maintained for years after vendors stop supporting them. The Mirai botnet of 2016 showed the scale of the problem: it took over hundreds of thousands of cameras and routers with default passwords and used them for large DDoS attacks.
- Default or shared credentials allow automated takeover; each device needs a unique secret.
- Unencrypted traffic exposes readings and commands; TLS or DTLS should protect every link.
- No update path leaves known flaws open for the device lifetime.
- Flat networks let one compromised device reach critical systems; segmentation limits damage.
- Excessive data collection creates privacy and compliance exposure without operational value.
Regulators have responded with baseline requirements. ETSI EN 303 645 sets security provisions for consumer IoT, and several jurisdictions now ban universal default passwords and require a published vulnerability disclosure policy and a stated support period.
Put simply
The Internet of Things means everyday objects that can talk over the Internet. A thermostat, a water meter or a factory motor gets a tiny computer and a radio, so it can report what it measures and receive instructions.
Think of it as giving objects a voice and ears. The smartphone in your pocket is used by a person, while an IoT device mostly works on its own, sending short status messages and reacting to commands. The benefit is automation and better information; the price is that every connected object must be kept secure, like a door that also needs a lock.
In practice
A municipal water utility serving about 120,000 households replaces manual meter reading with smart water meters on NB-IoT. Each meter sends one compact reading every hour and an immediate alert when flow never drops to zero for 24 hours, which signals a likely leak inside a home.
The meters are designed for a battery life of about twelve years and receive signed firmware updates over the air. Readings flow through the mobile operator network to a cloud platform that compares district inflow with the sum of household consumption.
Within the first year, the utility locates distribution leaks much faster because gaps between inflow and billed use show up per district within hours, and customers receive leak alerts days before a monthly bill would have revealed the problem.