The four main types of embedded systems
Embedded systems are most often grouped into four types based on how they work and what they need to do:
| Type | What defines it | Examples |
|---|---|---|
| Standalone | Works on its own, without a host computer or network | Microwave oven, washing machine, digital watch, calculator |
| Real-time | Must respond within a defined time limit | Airbag controller, anti-lock brakes, pacemaker, industrial motor drive |
| Networked | Relies on a wired or wireless network to do its job | IP security camera, smart thermostat, ATM, card payment terminal |
| Mobile | Portable, battery-powered, small | Fitness tracker, smartwatch, handheld barcode scanner, digital camera |
This classification is a teaching tool, not a strict set of boxes. Many real devices belong to two or three types at once. We’ll come back to that after looking at each type. If you’re new to the topic, start with our overview of what an embedded system is.
Standalone embedded systems
A standalone embedded system does its job independently. It takes inputs, processes them, and produces outputs without needing a PC, a server, or a network connection.
Classic examples include:
- Microwave ovens: reads the keypad, runs the timer, controls the magnetron and turntable, and updates the display
- Washing machines: sequences water fill, agitation, spin, and drain based on the chosen program and sensor readings
- Digital watches and calculators
- Basic MP3 players and electronic toys
- Simple temperature controllers for ovens, incubators, or aquariums
Standalone systems are often the simplest and cheapest kind of embedded system, frequently built around a single small microcontroller. Note that a standalone device can still have real-time requirements. A microwave’s door-interlock logic, for example, must react reliably.
Real-time embedded systems
A real-time embedded system must produce the correct result within a specific time. A correct answer delivered too late counts as a failure. “Real-time” doesn’t necessarily mean fast. It means predictable: the system must meet its deadlines consistently.
Real-time systems are usually divided into three categories.
Hard real-time
Missing a deadline counts as a system failure, potentially with serious consequences. Examples:
- Airbag deployment controllers
- Anti-lock braking and stability control
- Flight control systems
- Pacemakers and some other medical devices
- Industrial safety interlocks
Hard real-time systems are designed and tested so that their worst-case response time is known and always within limits.
Firm real-time
An occasional missed deadline doesn’t cause disaster, but a late result is useless and is thrown away. A classic example is a video or audio stream where a late frame is simply dropped, or an automated inspection system that skips an item it couldn’t analyze in time.
Soft real-time
Missing deadlines degrades quality but doesn’t cause failure, and a late result still has some value. Examples include:
- Streaming media players
- Video game consoles
- Smart TV interfaces
- Many consumer gadgets, where an occasional stutter is annoying but harmless
Real-time systems often use a real-time operating system (RTOS), which schedules tasks by priority so the most time-critical work always runs first. Simpler real-time systems may run bare-metal code driven by hardware interrupts.
Networked embedded systems
A networked embedded system is connected to a network (wired or wireless, local or internet) and relies on that connection to do its job or deliver its value.
Examples:
- IP security cameras streaming video to a recorder or cloud service
- Smart home devices such as thermostats, plugs, door locks, and lighting
- ATMs that verify transactions with a bank’s servers
- Point-of-sale and card payment terminals
- Industrial sensors and controllers on fieldbus or Ethernet networks
- Vehicle ECUs, which talk to each other over CAN and other in-car networks
Most Internet of Things (IoT) devices fall into this category. Networked systems bring extra design concerns: communication protocols, handling dropped connections, remote updates, and above all security, since a connected device can be attacked remotely.
Mobile embedded systems
Mobile embedded systems are portable devices designed to be carried or worn. They’re defined by small size, battery power, and tight limits on energy and memory.
Mobile embedded systems examples:
- Fitness trackers and smartwatches
- Wireless earbuds
- Digital cameras and action cameras
- Handheld barcode scanners and inventory terminals
- Portable GPS units
- Handheld medical devices such as blood glucose meters and pulse oximeters
- E-readers
Smartphones are often listed here too, though a modern smartphone is closer to a general-purpose computer in many ways. It runs apps the user chooses. Inside, though, it contains many embedded subsystems, such as the modem, power management, camera processing, and sensor hubs.
The defining challenge for mobile devices is power. Designers rely on low-power microcontrollers, aggressive sleep modes, and efficient radios like Bluetooth Low Energy to stretch battery life.
Most devices belong to more than one type
The four categories overlap constantly. Take a smartwatch:
- It’s mobile: worn on the wrist, battery-powered.
- It’s networked: syncs with a phone over Bluetooth, sometimes Wi-Fi or cellular.
- Parts of it are real-time: heart-rate sampling, touch response, and display updates have timing requirements, mostly soft.
- Some functions work standalone: it can keep time and track steps with no connection at all.
A car’s ECU is both real-time and networked. A connected insulin pump is mobile, networked, and has hard real-time elements. So it’s more useful to treat the types as properties a design may have than as exclusive labels.
Classifying embedded systems by complexity
Another common classification is by scale and complexity:
- Small-scale embedded systems use a single 8-bit or 16-bit microcontroller with limited memory, often powered by a battery. They run simple firmware, usually without an operating system. Examples: TV remotes, toys, simple sensors.
- Medium-scale embedded systems use 16-bit or 32-bit microcontrollers, sometimes several, with more memory and peripherals. They often run an RTOS and involve more complex software and communication. Examples: drone flight controllers, smart thermostats, medical monitors.
- Sophisticated (large-scale) embedded systems use powerful processors, multiple cores, or custom hardware such as FPGAs or ASICs. They may run embedded Linux or a safety-certified RTOS, and they tackle demanding jobs. Examples: car infotainment and driver-assistance systems, network routers and switches, avionics.
These boundaries are fuzzy. Today’s low-cost 32-bit microcontrollers have blurred the line between small and medium scale.
Other useful ways to classify embedded systems
Engineers often group embedded systems by the technology inside rather than by function:
By processor type:
- Microcontroller-based: most small and mid-sized devices. See microcontroller vs microprocessor for the difference.
- Microprocessor or SoC-based: devices that need more processing power, graphics, or a full OS
- FPGA or ASIC-based: devices that need custom hardware speed or very high volumes
By software stack:
- Bare metal: a main loop and interrupt handlers, with no operating system
- RTOS-based: a small real-time kernel such as FreeRTOS or Zephyr managing multiple tasks
- Embedded Linux (or similar): a full operating system on a capable processor, used in routers, infotainment, and industrial gateways
In day-to-day engineering, these categories often say more about how a system will be built than the four functional types do.
Frequently asked questions
What are the four types of embedded systems?
Standalone, real-time, networked, and mobile. Standalone systems work independently. Real-time systems must meet timing deadlines. Networked systems depend on a network connection. Mobile systems are portable and battery-powered.
What is embedded software?
Embedded software is the code that runs on an embedded system and controls its hardware. It’s usually written for one specific device, stored in non-volatile memory, and often called firmware. It ranges from a few kilobytes of bare-metal code to full embedded Linux systems. See firmware vs software for more.
Are embedded systems programmed in C or C++?
Both. C is the traditional and still most widespread language, especially on small microcontrollers. C++ is common on more capable chips and in larger projects. Rust, MicroPython, and other languages are used in some areas, but C and C++ dominate across all four types.