IoT and Edge AI

Embedded Systems and the Internet of Things

How embedded systems and the Internet of Things relate, where they differ, what's inside a connected device, and the trade-offs behind real IoT products.

On this page
  1. The short answer: how embedded systems and IoT relate
  2. What is an embedded system?
  3. What makes a device part of the Internet of Things?
  4. Embedded systems vs IoT: key differences
  5. Anatomy of an IoT embedded device
  6. What changes when an embedded system goes online
  7. Examples of embedded IoT devices
  8. Going deeper on embedded IoT
  9. Does IoT have a future?
  10. FAQ

The short answer: how embedded systems and IoT relate

An embedded system is a small, dedicated computer built into a larger device to do a specific job, like timing a washing machine cycle or controlling a car’s anti-lock brakes. The Internet of Things (IoT) is the idea of connecting physical devices to networks so they can send data, receive commands and work with other systems.

The two overlap heavily, and the relationship fits in one sentence: every IoT device contains an embedded system, but not every embedded system is part of the IoT. A digital kitchen scale is an embedded system. Give it a radio, a network stack and a cloud service that logs your recipes, and the same kind of embedded system is now an IoT device.

The rest of this guide explains each side, compares them directly, and walks through what actually changes in the hardware and firmware when an embedded system goes online.

What is an embedded system?

An embedded system combines hardware (usually a microcontroller or microprocessor, memory, and input/output peripherals) with software written for one purpose. Unlike a laptop, it isn’t meant to run whatever program you install. Its firmware is designed around the device it lives in.

Typical traits include:

  • Dedicated function. A thermostat controller regulates temperature. That’s its job.
  • Tight resource limits. Many microcontrollers have kilobytes to a few megabytes of flash and RAM, not gigabytes.
  • Interaction with the physical world. Sensors measure things and actuators move or switch things.
  • Real-time behaviour. Many embedded systems have to respond within a known time, such as firing an airbag or stepping a motor.
  • Long, unattended operation. Devices often run for years without anyone rebooting them.

The four common types of embedded systems

Textbooks commonly group embedded systems into four types by how they operate:

  1. Standalone embedded systems work on their own without a host computer or network, for example a microwave oven controller or a digital watch.
  2. Real-time embedded systems have to meet timing deadlines. They’re often split into hard real-time, where a missed deadline counts as a failure (engine control, airbag deployment), and soft real-time, where an occasional late response only lowers quality (audio streaming).
  3. Networked embedded systems connect to a local network or the internet to share data. Most IoT devices fall here.
  4. Mobile embedded systems are portable and battery-powered, such as fitness trackers, handheld scanners and phones.

These categories overlap. A smartwatch is mobile, networked and has real-time parts, all at once.

What makes a device part of the Internet of Things?

The IoT isn’t a single technology. It’s a pattern: physical devices with sensors and/or actuators, a network connection, and software somewhere else (a phone app, a gateway, a cloud service) that collects data, makes decisions or lets people control the device remotely.

A device usually counts as an IoT device when it:

  • Communicates over Wi-Fi, Bluetooth, Zigbee, Thread, LoRaWAN, cellular or another network.
  • Exchanges data with other systems, not just within itself.
  • Can be monitored or controlled remotely, often through a dashboard or app.
  • Fits into a larger system of devices, gateways, servers and users.

That larger system is what people mean by IoT architecture. We cover it layer by layer in IoT System Architecture: Layers and Components.

Embedded systems vs IoT: key differences

Aspect Embedded system IoT system
What it is A dedicated computer inside a device A network of devices plus the software and services around them
Scope One device Many devices, gateways, cloud and apps
Connectivity Optional, and often none Essential
Where data is processed Locally on the chip Split between device, edge and cloud
Main design priorities Reliability, timing, cost, power Plus networking, security, scalability, data management
Lifetime concerns Hardware wear, firmware bugs Plus security patches, cloud dependencies, protocol changes
Typical skills C/C++, electronics, RTOS, debugging Plus networking, protocols, cloud, security
Example A basic pacemaker A pacemaker that sends readings to a clinic’s monitoring system

The pacemaker example shows up in many explanations because it makes the point well. The medical device itself is an embedded system either way. Connecting it adds an IoT layer on top: radio links, data services and remote monitoring. It also adds security responsibilities that a standalone device never had.

Anatomy of an IoT embedded device

Most connected devices, from a smart plug to an industrial vibration sensor, are built from the same five building blocks.

Microcontroller or processor

Simple sensor nodes usually run on a microcontroller (MCU) such as an Arm Cortex-M part, an ESP32 or an nRF52/nRF54 series chip. These run firmware directly or on a small real-time operating system. Devices that need a display, a camera pipeline or a full TCP/IP feature set may use a microprocessor (MPU) running embedded Linux instead.

Many modern IoT chips are wireless SoCs with the radio and the processor on the same piece of silicon. That’s a big reason the ESP32 and Nordic’s nRF parts are so common in IoT designs.

Sensors and actuators

Sensors turn physical quantities (temperature, humidity, motion, light, current, gas concentration) into electrical signals, either analog through an ADC or digital over I2C, SPI or UART. Actuators work in the other direction: relays, motors, valves, LEDs and buzzers.

Radio and connectivity

The radio decides most of the device’s character: range, data rate, power draw and whether it needs a gateway. Wi-Fi gives direct internet access but uses a lot of power. Bluetooth LE is frugal but short-range. LoRa reaches kilometres at very low data rates. Zigbee and Thread form low-power meshes. For a detailed comparison, see Wi-Fi vs Bluetooth LE vs LoRa vs Zigbee.

Power

Mains-powered devices can keep the radio on all the time. Battery-powered devices usually spend most of their life in deep sleep, wake briefly to measure and transmit, then go back to sleep. Power design often decides more about the architecture than the processor choice does.

Firmware stack

A connected device’s firmware usually stacks up like this:

  • Hardware drivers and a hardware abstraction layer
  • A scheduler or RTOS (FreeRTOS, Zephyr and others), or a bare-metal loop for very simple devices
  • A network stack (TCP/IP, Bluetooth host stack, LoRaWAN MAC, or similar)
  • Security services: TLS, key storage, secure boot
  • An application protocol such as MQTT, CoAP or HTTP
  • An update mechanism for over-the-air firmware updates
  • The application logic itself

What changes when an embedded system goes online

Adding a radio sounds like a small change, but it touches almost every engineering decision.

Networking and protocols

The device now needs a network stack, connection management (reconnecting after Wi-Fi drops, for example) and an application protocol. MQTT, a lightweight publish/subscribe protocol, is a common choice for sending telemetry to a broker. Our ESP32 MQTT tutorial shows what that looks like in practice.

Security becomes a lifetime job

A standalone device can only be attacked by someone holding it. A connected device can potentially be reached by anyone on the network it joins, and sometimes from the internet. That means unique credentials instead of default passwords, encrypted links, protected keys, secure boot, and a plan for fixing vulnerabilities after launch. Regulations increasingly expect this. The EU’s Cyber Resilience Act and the UK’s product security rules, for instance, place obligations on makers of connected products.

When a connected device controls physical machinery, the stakes get higher still. That’s the territory of cyber-physical security.

Updates in the field

Once devices are deployed, you need a safe way to change their firmware remotely: fixing bugs, patching vulnerabilities, adding features. A failed update can brick a device that nobody can physically reach, so update design deserves care from the start. See Over-the-Air (OTA) Firmware Updates for IoT Devices.

Power budgets get harder

Radios are often the biggest power consumer in a small device. A sensor that could run for years on a coin cell as a standalone gadget may last weeks if it holds a Wi-Fi connection open. Connected designs lean heavily on sleep modes, batching data and choosing a low-power radio.

Examples of embedded IoT devices

Here are ten common devices where an embedded system sits at the heart of an IoT product:

  1. Smart thermostats that learn schedules and can be adjusted from a phone
  2. Smart plugs and switches that report energy use and accept remote commands
  3. Video doorbells and security cameras that stream to an app and detect motion
  4. Fitness trackers and smartwatches that sync health data through a phone
  5. Smart meters that report electricity, gas or water use to the utility
  6. Soil moisture and weather sensors in agriculture, often using LoRaWAN
  7. Asset trackers on shipping containers and pallets, often using cellular IoT
  8. Industrial vibration and temperature sensors for predictive maintenance
  9. Connected medical devices such as remote patient monitors
  10. Smart locks that use Bluetooth or Thread to unlock from a phone

If you want to build one yourself, our list of IoT projects using embedded systems covers hobby-scale versions of several of these.

Going deeper on embedded IoT

Each of these topics has its own article in this cluster:

Does IoT have a future?

Connected devices are already routine in homes, factories, farms, hospitals and cities, and the underlying parts (cheap wireless microcontrollers, low-power radios, cloud platforms) keep getting more capable and cheaper. Growth forecasts differ widely between analysts, so treat any single number with caution. The overall direction is fairly clear, though: more devices are being designed with connectivity built in from the start.

The field is also maturing. Recent priorities include stronger security, interoperability standards such as Matter for smart homes, and more processing on the device itself (edge AI) instead of sending everything to the cloud. For engineers, that means embedded skills paired with networking, security and some data knowledge stay useful.

FAQ

Is IoT a type of embedded system? Not exactly. IoT describes a whole connected system: devices, networks, services and users. The devices inside that system are built around embedded systems. You could say IoT is built on top of embedded systems.

Can an embedded system work without the internet? Yes. Most embedded systems ever built have no internet connection at all. Car engine controllers, appliance controllers and many medical and industrial devices run standalone or on closed local networks.

Do I need to learn embedded systems before IoT? For hardware-side IoT work, yes, it helps a great deal. Understanding microcontrollers, peripherals, interrupts, power and C or C++ makes it much easier to build reliable connected devices. Cloud-side IoT work leans more on networking and backend skills.

Which microcontroller is good for learning IoT? The ESP32 family is a popular starting point because it combines a capable processor with Wi-Fi and Bluetooth, is inexpensive, and has large community support in both the Arduino ecosystem and Espressif’s ESP-IDF framework.

Keep exploring

Guides from other topics

All topics →