15.3 Embedded Systems, the Internet of Things, and Types of Computing Systems
Key Takeaways
- An embedded system is a computer built into a larger device to perform a dedicated function, often with limited resources and real-time requirements, as in ATMs, pacemakers, car braking systems, and microwave ovens.
- The Internet of Things (IoT) is the network of everyday physical objects with sensors, processors, and Internet connectivity that collect and exchange data, such as smart thermostats and connected refrigerators.
- Embedded systems use sensors to measure the environment, process the readings, and drive actuators in a sense–process–act loop.
- A cluster is a group of networked computers that work together as one system for high performance or high availability.
- Compare computing systems on processing power, memory and storage, portability, battery life, cost, energy use, reliability, scalability, and connectivity, relative to the task.
What this competency asks
Two ETS competencies are covered here:
- Be familiar with computing systems embedded in everyday objects (for example, IoT, ATMs, and medical devices): describe what an embedded system is, define what the IoT is and how it is used, and describe how sensors are used in embedded systems.
- Know the capabilities, features, and uses of different types of computing systems (for example, desktop, mobile, and cluster): identify them for each type, and identify criteria to evaluate and compare computing systems.
ETS's sample question asks which scenario is an IoT example. The answer is a refrigerator that orders milk from an online service when supplies run low. A car's collision warning, a GPS watch, and a heart-rate watch are embedded computing, but they do not need the Internet to work.
Embedded systems
An embedded system is a computer that is part of a larger device and is dedicated to specific functions.
| Characteristic | Explanation |
|---|---|
| Dedicated purpose | Runs one program, or a few, for the device's job |
| Microcontroller | A small chip combining CPU, memory, and input/output on one chip |
| Limited resources | Little memory, low power use, low cost |
| Real-time | Often must respond within strict time limits, as with airbags and anti-lock brakes |
| Reliability | Must run unattended for years; failures can be dangerous |
| Little or no conventional interface | Buttons, a small display, or none at all |
Examples: ATMs, pacemakers and insulin pumps, car engine controllers, anti-lock brakes, airbag systems, microwave ovens, washing machines, digital cameras, traffic signals, and elevator controllers. There are far more embedded computers in the world than desktop and laptop computers combined.
Sensors in embedded systems
Embedded systems interact with the physical world through a sense → process → act loop:
- Sensors measure a physical quantity such as temperature, light, motion, pressure, sound, or location. An analog-to-digital converter turns the measurement into a number.
- The processor compares the reading with thresholds or runs a control algorithm.
- Actuators act: motors, heaters, valves, alarms, displays.
- The loop repeats, often many times per second. This is feedback control.
Example: a thermostat. Read the temperature every few seconds. If it is below the set point, turn the heat on. If it is above the set point plus a small margin, turn the heat off. The margin prevents rapid on-off switching.
Design issues include sampling rate (how often to read), sensor noise and calibration, power consumption (reading less often saves battery), and safety when a sensor fails.
The Internet of Things
The Internet of Things extends embedded systems with network connectivity, so everyday objects collect data, communicate over the Internet, and can be monitored or controlled remotely.
| Area | IoT example |
|---|---|
| Home | Smart thermostats, doorbells with cameras, voice assistants, connected appliances |
| Health | Remote patient monitoring, connected glucose monitors |
| Agriculture | Soil-moisture sensors that trigger irrigation |
| Cities | Smart parking, adaptive traffic signals, connected streetlights |
| Industry | Machines that report their condition before they fail (predictive maintenance) |
| Schools | Energy management, occupancy sensors, asset tracking |
Benefits: automation, remote control, data for better decisions, energy savings, convenience.
Concerns:
- Security: many devices ship with weak default passwords and are rarely updated. In 2016 the Mirai botnet took over large numbers of IoT devices, such as cameras and video recorders, and used them in massive denial-of-service attacks (Section 16.6).
- Privacy: cameras, microphones, and location sensors collect sensitive data continuously (Section 3.3).
- Dependence: a device may stop working when its manufacturer's cloud service shuts down.
- Interoperability among brands, and the energy and waste of many small devices.
Types of computing systems
| System | Capabilities and features | Typical uses |
|---|---|---|
| Desktop | Powerful CPU and GPU; large screen; expandable; plugged in | Programming, design, labs, gaming |
| Laptop | Desktop-class software; portable; battery powered | Students and teachers on the move |
| Mobile (smartphone, tablet) | Very portable; touch; cameras and sensors (GPS, accelerometer); cellular; limited input for heavy work | Communication, field data collection, apps |
| Wearable | Tiny; body sensors; long battery life | Fitness and health monitoring |
| Server | Reliable, always on, many users; often "headless" (no monitor) | Websites, email, databases, file storage |
| Cluster | Many networked computers acting as one; scales out; tolerates failure of individual machines | Web services with heavy traffic; scientific computing; big data |
| Supercomputer | Enormous parallel computing power | Climate modeling, physics, drug discovery |
| Mainframe | Extremely reliable; massive transaction volume | Banks, airlines, government records |
| Embedded | Dedicated, low-power, often real-time | Appliances, vehicles, medical devices |
A cluster divides work among its nodes (parallel computing) and keeps running if one node fails (high availability). Cloud providers run huge clusters in data centers (Section 15.4).
Criteria for evaluating and comparing systems
| Criterion | Questions to ask |
|---|---|
| Processing power | Number of cores, clock speed, GPU: enough for the workload? |
| Memory and storage | Enough RAM for the software; enough storage, and how fast? |
| Portability and size | Must it be carried into the field or classroom? |
| Battery life and energy use | Unplugged use; operating cost; heat |
| Cost | Purchase price plus software, maintenance, and support |
| Reliability and availability | What happens if it fails? Is redundancy needed? |
| Scalability | Can it grow with more users or data? |
| Connectivity | Wi-Fi, cellular, wired; which ports? |
| Input and output | Keyboard for coding? Touch? Sensors? Screen size? |
| Security and manageability | Can it be updated, locked down, and managed centrally? |
| Software compatibility | Does it run the required applications? |
The best system depends on the task. A cluster is wasted on word processing, and a phone is a poor choice for compiling a large project. The best answer usually matches the system's strengths to the stated requirements.
Which scenario is the clearest example of an Internet of Things device?
Which description best defines an embedded system?
A popular website must stay available even if individual machines fail, and it must handle rapidly growing traffic. Which computing system best fits these requirements?
Students will record GPS locations and photos of plants on a field trip and upload them from the field. Which criteria matter most when choosing a device?