LLC NEWLAND technology / 03.09.2020

What are the roles of sensors in an IoT network?

The Internet of Things (IoT) and edge computing have created many smart industries such as smart city, smart factories, smart agriculture, smart medicine, and more. How do they become smart? The key lies in real-time data capture and analytics, using various types of sensors.

TechJury, a software firm that reviews forecasts from various companies, predicts that IoT connections will reach 64 billion by 2025, while Grand View Research, a marketing firm, projects the IoT market will reach $949.42 billion the same year. The accelerating growth of the sensor industry is expected to mirror that of the IoT. Sensors are innovating very fast. It is expected that in the next 25 years, they will be smaller, smarter, cheaper, and ubiquitous.

How do sensors work, and what is involved in IoT system development using sensors? This series will begin answering some common questions raised by many.

Q. How do sensors work?

A. IoT cloud servers and edge (gateway) devices depend on sensors to collect real-time data. Since the world we live in deals with analog signals such as temperature in Fahrenheit, distance in feet, speed in miles per hour, pressure in pounds per square inch, etc., sensors have to detect changes in their environments and then convert the physical parameters’ signals to digital data.

Figure 1 illustrates the configuration of an IoT network with sensors. Sensors are usually referred to as the endpoint in the IoT network because they are at the outermost edge from the cloud servers’ perspective. Note the cloud servers on the right side of Figure 1. Even though these sensors are small and do not seem as important as the cloud servers, they can play a critical role in system design. A recent notable example comes from the Boeing 737 Max plane crashes. Among other factors, sensor failure played a part in the two tragedies.

Sensors need to connect and communicate with the gateways and cloud servers to operate in the IoT network. Nowadays, wireless technologies such as Bluetooth, NFC, RF, Wi-Fi, LoRaWAN, NB-IoT (cell) are commonly used. Depending on how the network is set up, more and more gateway products are used to perform edge computing (data analytics) before sending data to the cloud servers.

Q. How many different types of sensors are available today?

A. There are hundreds of different types of sensors available today to measure almost anything, and they come in various forms and shapes. Some are a single component such as a light-sensing diode; others can be a module with built-in microcontrollers.

These sensors are available to measure light, sound, temperature, pressure, position, change of altitude and distance, types of gases, movement, liquid and on and on. There are also many types of technologies used for detection including radar, LiDAR, light detection, magnet detection, infrared (IR), using inductive technology, image sensing, ultrasonic, sonar, photon emission, touch sensing, encoder and many more. Here is a sample list of the types of sensors available.

  • Atmospheric pressure sensor
  • Distance/ Proximity sensor
  • Humidity sensor
  • IR sensor
  • Level sensor
  • Light sensor
  • Motion detection sensor
  • Pressure sensor
  • Smoke and gas sensor
  • Temperature and thermocouple sensor
  • Touch sensor
  • Ultrasonic sensor

Additionally, sensors can also be classified into active vs. passive and analog vs. digital.

Active sensors typically require external stimulation to work. It can be an external power source or induced energy. For example, a linear variable differential transformer (LVDT) can be used to convert linear motion to equivalent electrical signals. In this case, the energy source comes from the linear movement through induction, even without an external power source. Passive sensors do not need external stimulation to work. For example, a thermocouple will be able to convert heat directly to electrical signals.

Both analog and digital sensors are used. Analog sensors measure analog signals such as temperature and pressure. These output signals need to be digitized before communication with microcontrollers can take place. Digital sensors, on the other hand, can measure input signals such as light and use an internal rotating disc, for example, to output in digital formats.

Q. What are the considerations in choosing a sensor solution?

A. A few criteria should be considered when selecting a sensor solution:

Accuracy in terms of precision and resolution. Determine in advance what resolution you need. In the case of temperature, will accuracy within 1 degree be accurate enough? For distance, you may need accuracy as low as units of 0.01 inch. The higher the resolution, the higher the cost. Measurement range. What is the upper and lower limit you need? If you don’t need a wide range, it will reduce your design cost.

Environmental conditions. Does it have to perform in high temperature/humidity conditions or be waterproof?

Reliability considerations. There are two aspects. One is how long the sensors will function without breaking down. The mean-time-between-failure (MTBF) specification refers to the length of time that sensors can function without experiencing failure or breakdown. This will be important for sensors operating in remote and rugged areas. The other aspect is whether the sensor’s performance is accurate without calibration. It is possible to have sensors that have not broken down but require calibration to perform accurately.

Rugged environment. Some sensors must withstand extreme heat, cold, shock, and vibration. Make sure the sensor you select will meet the requirements.

Size and weight requirements. Depending on the applications, some sensors must fit in very small spaces, or many sensors need to be installed for one application.

Tradeoffs—highly integrated solution vs. using individual components. An integrated solution may cost a bit more, but can ave design and installation time. For example, a vibration sensor with a built-in LoRaWAN or RF communication function will take up less space and installation time.

If the applications are critical and require redundancy, consider sensors with a dual-package. Overall system cost. Finally, take a look at the overall design, install, and support costs and not only the sensor’s individual unit cost.