Measurement aspects when selecting sensors for your IoT application
Blog - Measurement-Specialists

What are the most important criteria to consider when choosing a sensor? In this post I will focus on some of these – the ones that are directly related to the measurement. Other aspects will be discussed in a later post.

Have you already read my post The process of selecting and evaluating IoT sensors? Keep then in mind that the measurement aspects discussed in this post are important throughout the process of selecting sensors, from defining the requirements, to evaluating measurements from your prototype.


The resolution of your measurement is the smallest difference that is possible to measure. Let’s look at an example where you want to measure the distance to an object that is 2 meters away. If the resolution of the sensor is 3 mm, then the values that you get can be 1.997 m, 2.000 m and 2.003 m. Only looking at the measurement values here one might think that the resolution is 1 mm. This is something to be aware of.

Measurement range

Only values measured within the measurement range can be considered valid. Therefore it’s important to choose a sensor that covers all possible values that you want to measure. For example, returning to the sensor measuring distance, it might have a measurement range between 1 m and 3 m. Then objects closer to the sensor than 1 m, as well as objects further away than 3 m, will produce incorrect signal values.

Sensitivity and noise

It’s hard to talk about noise if not set in relation to the sensitivity of a sensor. Sensitivity is the ratio between the output signal and the value of the measured property. For example if measuring a distance to an object with a sensor that has an voltage output, the sensitivity is given in V/m.

Noise is unwanted modifications of the measured signal. For a sensor it is often given in signal to noise ratio (SNR). So by knowing the signal strength and the SNR in the range you want to measure you can find the sensor’s noise contribution to the accuracy. In the final system the sensor’s noise is often dependent on e.g. electrical interference, vibrations, temperature. These parameters need to be considered to get an overall view of the total noise.


The accuracy is how well the value measured conforms to the actual value. The accuracy is a combination of several factors. The resolutions can be a limiting factor for how close to the actual value the measurement is, but it is not very common. One application where that might be the case is in vision systems where the pixel size of a camera might be limiting the accuracy. In many cases sensitivity and noise plays a bigger role.

All sources of inaccuracy (resolution, noise, systematic errors and ageing) have to be added up when making an estimation of the accuracy in your final application. An accuracy analysis can be done later on by making measurements using the sensor on a known reference, but a good estimation is important for making the right choice of sensor.

Repeatability and precision

It’s important that the measurements of a given property, the length of an object for instance, do not vary over time, both between measurements taken directly after each other as well as between measurements taken far apart in time. This can be referred to as repeatability or precision, and is closely related to accuracy.

Variations over time can be due to environmental aspects as described above, but some sensors also age, and then it’s important to know how to compensate for that, especially if it is not possible to perform regular calibrations. Perhaps another sensor ages less.

Sample rate

Sample rate, or sampling frequency, is how often measurement values are taken. It is related to the nature of what you want to measure and how it is going to be analysed. As stated by the Nyquist criteria, the sample rate should be at least twice the frequency of the variations in the measured signal. It is therefore important that the sensor supports a high enough sample rate.

A higher sample rate can sometimes be used to reduce noise by applying filters to the signal afterwards, still keeping the Nyquist criteria in mind. However, sampling too often also comes with drawbacks since it will make the IoT device consume more energy and it will increase the bandwidth for data transmission. Both the sensor itself, the data processing on the device and the amount of data transferred to the cloud requires more energy with higher sampling rate.

Selecting your sensors with care

I hope this gave you a good overview of parameters that are important to think of when selecting sensors. Next post will add more aspects – the once not related to the measurement itself.

By Karin Hellqvist

Linus Amoli