The short answer
There’s no official, standardised list. “Seven components” is a common way of breaking the topic down, and different vendors divide it slightly differently. A widely used technical breakdown of an IoT platform is:
- Devices: sensors, actuators and the embedded firmware that runs them
- Connectivity: the networks and protocols that move data
- Gateways and edge computing: local hubs that bridge, filter and process data
- Cloud platform core: message ingestion, device identity and device management
- Data storage, processing and analytics: databases, rules engines and machine learning
- Applications and user interfaces: dashboards, mobile apps, alerts and APIs
- Security: protections that run through every other component
Some lists, especially ones describing the wider IoT ecosystem, swap a couple of these for users and use cases. We cover that version further down.
The seven components explained
1. Devices: sensors, actuators and embedded firmware
Devices are the “things” in the Internet of Things. Each one is built around an embedded system: a microcontroller or small processor running firmware that reads sensors (temperature, motion, current, location, images) and drives actuators (relays, motors, valves, locks, LEDs).
The firmware on the device does more than read sensors. It manages power and sleep, handles the network connection, buffers data when the link drops, applies local logic, and accepts updates. Some lists treat this on-device software as a separate component called the “application in the device.” For more on what’s inside a connected device, see Embedded Systems and the Internet of Things.
2. Connectivity
Connectivity carries data between devices, gateways and the cloud. It has two parts:
- The link: Wi-Fi, Bluetooth LE, Zigbee, Thread, LoRaWAN, NB-IoT, LTE-M, Ethernet, or industrial buses
- The messaging protocol: MQTT, CoAP, HTTP(S), LwM2M and others
The right link depends on range, data rate, power budget and cost. Our comparison of Wi-Fi vs Bluetooth LE vs LoRa vs Zigbee goes into the trade-offs, and the ESP32 MQTT tutorial shows a messaging protocol in action.
3. Gateways and edge computing
Many devices can’t reach the internet directly. A Zigbee sensor or a LoRa node needs a gateway to bridge its radio to an IP network. Gateways often also:
- Translate protocols (for example Modbus or Zigbee to MQTT)
- Filter and aggregate data before sending it upstream
- Keep local automations running when the internet is down
- Run edge computing workloads, such as machine learning models that analyse data close to where it’s produced
Devices that connect directly over Wi-Fi or cellular may skip a separate gateway, but the edge-processing role often still exists on the device itself.
4. Cloud platform core: ingestion and device management
This is the heart of what vendors usually mean by an “IoT platform.” It typically provides:
- Ingestion: a message broker or IoT hub that accepts connections from large numbers of devices
- Device registry and identity: a record of every device, its credentials and its metadata
- Device management: provisioning new devices, configuration, health monitoring, remote commands, and over-the-air firmware updates
- Device state: a stored copy of each device’s reported and desired state (sometimes called a device shadow or device twin), so apps can work with devices that are temporarily offline
5. Data storage, processing and analytics
Raw readings become useful only after they’re stored and processed:
- Storage: time-series databases for sensor data, plus data lakes or warehouses for long-term analysis
- Rules engines: simple logic like “if temperature is above 30°C for five minutes, raise an alert”
- Stream processing for real-time calculations on incoming data
- Analytics and machine learning for trends, anomaly detection and predictive maintenance
6. Applications and user interfaces
This is what people actually see and use: web dashboards, mobile apps, notifications, reports, and APIs that feed other business systems such as maintenance, billing or ERP software. It’s also where many commands start, like a user tapping “unlock” or a scheduled rule switching on irrigation.
7. Security
Security isn’t really one box. It runs through every other component:
- Devices: secure boot, protected key storage, unique credentials, locked debug ports
- Connectivity: encrypted, authenticated links (TLS, DTLS, or the link-layer security built into BLE, Zigbee and LoRaWAN)
- Cloud: per-device identities, least-privilege access rules, audit logs
- Applications: user authentication and authorisation
- Lifecycle: patching vulnerabilities through secure updates, and decommissioning devices safely
When devices control physical equipment, security failures can have physical consequences. See What is cyber-physical security?
The “ecosystem” version of the list
Some sources describe the seven components of the IoT ecosystem instead of the platform. A commonly cited version, used by connectivity provider Hologram among others, is:
- IoT devices
- Device connectivity
- The application running on the smart device
- Security
- IoT gateway
- Applications (the use cases the devices serve)
- IoT users
This list is less about backend software and more about everything that has to come together for an IoT product to succeed, including the people using it. Both lists are reasonable. If you’re studying for an exam, use whichever version your course material uses.
How the components map to architecture layers
| Component | Typical architecture layer |
|---|---|
| Devices | Perception (device) layer |
| Connectivity | Network layer |
| Gateways and edge | Edge / gateway layer |
| Cloud platform core | Processing / middleware layer |
| Data and analytics | Processing / middleware layer |
| Applications and UI | Application layer |
| Security | Cross-cutting, across all layers |
For more on the layer models themselves, see IoT System Architecture: Layers and Components.
Example: the seven components in a smart building
Consider an office building that monitors occupancy and air quality to control ventilation:
- Devices: battery-powered CO₂ and occupancy sensors in each room, plus actuators on ventilation dampers
- Connectivity: the sensors use a low-power mesh network such as Zigbee or Thread, while the gateways use the building’s Ethernet
- Gateways and edge: one gateway per floor collects readings and adjusts dampers locally, so ventilation keeps working if the internet connection fails
- Cloud core: a managed IoT service registers every sensor, tracks battery levels and pushes firmware updates
- Data and analytics: readings go into a time-series database. Analytics show which rooms are under-used and flag sensors that are drifting
- Applications: facilities staff use a dashboard and get alerts, and an API feeds room-booking software
- Security: each sensor has its own credentials, links are encrypted, the building network keeps IoT traffic separate from office traffic, and updates are signed
Build vs buy
Managed cloud services such as AWS IoT Core and Azure IoT Hub, along with many specialist IoT platforms, cover the cloud core and parts of data handling. Open-source options such as the Mosquitto or EMQX brokers, the Node-RED automation tool, ThingsBoard and time-series databases let teams self-host. Even with a managed platform, you still own the device firmware, connectivity choices, application and most of the security design. In practice, the device side is often where projects spend the most engineering effort.
FAQ
Is an IoT platform the same as IoT architecture? Not quite. Architecture is the overall structure of a connected system across all layers. An IoT platform usually means the software, often cloud-based, that connects, manages and processes data from devices. It covers components 4 and 5 most directly, with some vendors including parts of the others.
What is the most important component? It depends on the project, but many IoT failures trace back to devices (power, reliability, field updates) and security rather than the cloud software. A platform can be replaced later. Deployed hardware is much harder to change.
Do all IoT systems need a gateway? No. Devices with Wi-Fi, Ethernet or cellular connections can talk to the cloud directly. Gateways are needed when devices use radios like Zigbee, Thread, BLE or LoRaWAN that don’t connect to the internet on their own, or when local processing is required.
What are some examples of IoT devices? Smart thermostats, smart plugs, fitness trackers, connected cameras, smart meters and industrial vibration sensors are common examples. For hands-on ideas, see examples of IoT projects using embedded systems.