Since the data is an indispensable focus of IoT (Internet of Things), it is extremely important to ensure its accuracy. This is a brief overview of where data is generated and how sensors and actuators are an important factor in the precision and credibility of IoT data.
At the beginning or the lowest layer of IoT ecosystem, we find a world of things or, in other words, a perception layer. Here is a source of data, where the events of the physical, material world, with the help of technology, are transformed into signals and a digital format suitable for further transmission and processing within the IoT ecosystem. Organizers, modules or even hardware subsystems that make this happen are known as sensors. The name comes from the Latin word „sentire„ which means „to feel“. Sometimes the expression „detector“ is used because it is their role to „detect“ events or changes in their environment and to forward information to other components of the system. More simply, the sensor is in the IT system with the role of producing a digital signal, having the ability to communicate by signal transmission and to perform logic functions and instructions.
Considering the current perspective of the development of elements used in the compilation of the IoT ecosystem, there are several classifications of sensors.
One can basically distinguish between passive and active sensors. The first, i.e. passive, do not require an additional external power source for monitoring event environments while active ones must have a power source to operate.
Also, one of the classifications of the sensor is based on the method used for detection and measurement, that is we have mechanical, chemical, thermal, gas, IR, etc.
Also viewed with regard to the type of signal detected,there are analogue and digital sensors. The names themselves indicate that one group of sensors has been declared to have analogue signals in their functioning while the other group is based on digital/discrete signals.
Due to the dynamics and accelerated development of IoT components in different domains, a more recent classification is increasingly presenting on the level of independent functioning and possible decision-making inside a process. So we find the following classification:
– Base sensors – those that have a physical sensor without the addition of embedded processors for signal processing, but only for forwarding of the measured value. Some of the base sensors are Accelerometer, Gyroscope, Magnetometer, Proximity, Heartrate …
– Smart sensors – are those that incorporate some of the digital motion processors (DMPs). This means that this class of sensors takes the input signal as the value of the state of an event or the value of a physical process and then uses embedded computing, performs a certain calculation, processing and processing before the signal is transmitted to the network layer of the IoT ecosystem. At the same time, these sensors consist of at least one physical sensor, low-power microprocessor and communication module. In addition to these basic components, the smart sensor may also contain some of the IoT components such as compensation filters, amplifiers, transducers and other. Additionally also simultaneous software embedded elements that enable the functionality of a type as: conversion and digital data processing with the support of communication software along with the help of predefined IoT protocols.
– Intelligent sensors – as perhaps their name implies, this type of sensor is equipped with components so it can overwhelm a number of intelligent modalities such as self-validation, automated adaptation, and self-identification and also a plethora of self-testing functions. One of the basic differences with regard to smart sensors is that intelligent sensors can respond to certain activities after the detected changes. This also points to the fact that intelligent sensors have the ability to control responses based on stimuli of the external environment and even stimuli outside the IoT ecosystem. Advanced learning methods, as well as behavioral adaptations with the support of advanced processing signal, are integrated within the sensor itself. This declares this type of sensor into a special group of specialized hardware, but is currently highly anticipated in the design of the IoT ecosystem.
But what can be done with the data?
In the first thought, the data is further processed in the IoT ecosystem or outside the system. It can also be analyzed for the purpose of detecting anomalies or prediction as an example. Besides, it is possible to produce data based reports. However, data can be used as a trigger for another event. This is particularly interesting in the scenario where the data acts as activator of a given data pattern in some process.
Imagine the situation in which the water pump should be switched off and halt on irrigation of some area in the farming plantation. In this case, the design of the system would probably have looked like measuring the soil moisture sensors set on the plant. By configuring the system, the sensors are configured or in operation using some of the machine learning methods that if a certain level of soil moisture has been met, they then generate data that will send to some microcontroller which activates an event on the irrigation pumps and stop the water supply. In this case, the water pump is, in fact, the actuator in our IoT ecosystem.
By definition, the actuator is actually a device that transforms a certain form of energy into motion. As their mechanism converts energy into motion, we can categorize them based on energy sources:
– Pneumatic – use compressed air for generating motion.
– Hydraulic – use the liquid for generating motion.
– Thermal – use a heat source for generating motion.
– Electric – use external energy sources such as batteries or other types of electric energy to generate motion.
Observed by the motion patterns that actuators produce, there are those who create linear, oscillatory or rotating motion and thus act on some device or thing within the IoT ecosystem.
The most frequent use of actuator form is in manufacturing and assembling processes. For example, a type of impeller cylinder, which uses stored compressed air energy inside a metal cylinder to move the working piston when the air is released or not compressed.
Also in the robotics, various types of grippers are triggered by actuators with compressed air and act like human hands or fingers to perform some grabbing, drawing, pouring, granular jamming…
Putting all together
From the aforementioned, these components may be self-contained units or as part of the IoT ecosystem crucial building elements. Taking into account the previous example of irrigation control, an upgraded representation of the IoT ecosystem is shown in the following illustration.
Considering the design of the IoT system, taking into account the choice of the system architecture model, it is also necessary to guide the individual sensor and acoustic selection from the very beginning of the project. It is especially important to take into account the accuracy of the sensors that are implemented in the system. We can freely say that in terms of the correct functioning of an IoT system, the sensor accuracy is a critical factor.
Basically, accuracy is a simple description of how much the indicated value is close to, i.e. credibly represents the measured size that is observed in some process. All possible sources of faults that are relevant to the ecosystem, sensor type and type of measurement shall be taken into consideration.
The accuracy of the sensor has to be controlled timely – the process is called calibration. Sensor manufacturers are obliged to provide accurate data on wavelengths, percentages of measuring errors, as well as the sensor calibration procedures. The designers of the IoT system must comply with these manufacturer’s notes in order to ensure the proper functionality of the system.
Depending on the nature of the IoT system, the importance of data quality that comes from the layer of things within the system is indispensable. If the endpoint system is for example prediction or real-time analysis, and at the initial stage we have chosen low-accuracy sensors, it is logical that the ultimate product will be evaluated on loss performance.
There are also a number of parameters that affect the correct functioning of an IoT ecosystem and are directly related to sensors and their accuracy or level of error (Linearity, Repeatability, Resolution, Hysteresis …). Their details will be described in the following articles as parts of the description of specific IoT solutions.