Knowing the components of IoT helps us to understand what it’s all about. There are four major components of IoT:
Sensors and actuators: these components perceive information from and interact with the environment. Sensors collect data, while the actuators allow the system to perform specific actions. In the coffee machine example, the sensor monitors when the user wakes up, and an actuator starts making the coffee.
Connectivity: the sensors and actuators are connected to a so-called gateway. The gateway is responsible for communicating with the nearby sensors and actuators, translating the messages into a common format which is then uploaded to a cloud service on the internet. The communication between the gateway and the sensors and actuators is typically wireless, however wired connection is also possible. Cloud services are typically very cautious about privacy (if you would like to know more, you may want to check out our Digital revolution course and Cybersecurity course).
Cloud: the computer “cloud” – which is a network of computers on the internet – is responsible for storing and analysing the data in order to make smart decisions. The analysis might involve simple rules or complex artificial intelligence (AI) algorithms. You can find more information about AI in the Elements of AI course.
Human-Machine Interaction (HMI): data and analysis are supervised by users either on dedicated user interfaces (UIs) or smartphones and tablets with specific applications. The goal of HMI in an IoT system is to inform the user and to let them override the automated decisions if needed.
Sensors, actuators and user interfaces
The large number of possible sensors and actuators makes IoT systems very flexible. By using a different combination of sensors and actuators, systems can be tailored to fit different needs. In this section we’ll look more closely at the sensors, actuators and UIs that make up IoT.
There are two types of sensors: general-purpose sensors and task-specific sensors.
General-purpose sensors utilise common devices, such as cameras and microphones, to collect data. In this case the data, for example recorded images or sound, are analysed by computer applications and algorithms.
Task-specific sensors are developed to be cheap, energy efficient and/or robust while measuring specific values. Task-specific sensors typically require simpler software solutions than general-purpose sensors. Such sensors include thermometers, humidity meters, motion detectors, pulse and ECG monitors, and scale sensors.
Considering these two types of sensors, motion detection, for instance, can be realised in various ways:
a camera can be used to recognise a particular person or to cover movement in a larger area
a dedicated motion sensor might be utilised for cost and energy efficiency
a microphone can be used in hot, dark places, where the other two types of sensors would fail
Besides sensors, it’s also common to be able to interact with the machines involved in the IoT system, or with humans operating or using it. Therefore, actuators and user interfaces (UI) may be involved in IoT solutions. Actuators can change the values that are measured by the sensors, such as raising the temperature or opening the windows.
User interfaces are for human-machine interaction and they can be visual, audio or voice-enabled or haptic.
Visual UIs are mostly used with touchscreen smart devices (smartphones and tablets), or sometimes with a dedicated device with a touchscreen or a screen and buttons.
With a speech interface, the users interact with the IoT system using spoken natural language, while sounds can indicate important events with a pure audio UI.
Haptic interfaces usually use vibration to give feedback to the user.
A combination of the above UI types can improve the overall user experience.
IoT and 5G
5G is the fifth and most recent iteration of the mobile network. The first generation (1G) mobile network offered voice only. The novelty of 2G was digital networks, which is technically a much better solution than the analogue approach of 1G. 3G introduced data transfers, while 4G significantly increased the transmission speeds. 5G offers much faster data transfer than any previous mobile network, including 4G. It is also able to handle regions with higher connectivity density and provides better coverage.
A key element of 5G is Massive Multiple Input Multiple Output (MIMO), which involves a large number of antennas and complex communication software that makes fast communication with a large number of devices possible. As IoT depends on transferring and receiving data, high speed mobile internet is a major factor affecting the capabilities and reach of these smart devices. At the time of writing, it is thought that as 5G becomes widespread it will revolutionise IoT.
IoT and AI
When IoT meets AI, it's often referred to as AIoT (artificial intelligence/Internet of Things). Based on several sources, the benefit of IoT can be maximised with AI. To understand AIoT, it's important to understand the basics of AI. We have already learned that all the sensors and actuators collect data that is stored in the cloud. This data can be analysed by experts or by computer programs, written according to the experience and suggestions of experts. In cases where the data is complex, the rules written by the experts will not work perfectly. It’s also common that these rules are not general, so they need to be rewritten time after time as, for example, new sensors are added to the system or the environment changes (for example a new family moves into an IoT-enabled smart house). AI makes it possible that, with a high amount of representative data, the rules can be learned from the data itself – meaning the AI is also able to automatically adapt to changes.
In IoT, the integrator and decision-maker is always the human being, which is why the process is personalised and customised. With AI, devices use the data to learn, and they make decisions based on the previous behaviour of the user. More data results in better AI-based decisions. For instance, in the case of gender recognition, the AI has to be fed with photos of women and men. After seeing more and more photos, the AI can decide autonomously with better and better precision if a woman or a man is on the given image.
With AIoT, the large amount of data collected by the sensors combined with AI methods allows the system to learn both the users' behaviour and the environment and provide better services. Most AIoT solutions run on the cloud or around the gateway, and a novel approach brings AI to edge devices. The latter approach is referred to as TinyML. In this case the IoT devices may preserve their data locally, don’t need internet connectivity, and the response time can be much faster. On the other hand, TinyML algorithms face several constraints, like low energy consumption (as low as running a device for a year with a CR2032 small battery) and limited processing power.
For a detailed explanation about the mechanism and possibilities of AI you may want to look at the Elements of AI course.
IoT in our daily life
We already have sensors all around us – some of them are connected to a network, while some of them are just connected to a local machine. Smartwatches measure heart rate, daily activities and sleep times, and upload this data to the internet. The actual locations of buses and trams are collected based on GPS data and displayed on a map, and waiting times are displayed at stops. Smart cars have built-in sensors that collect and share data about wear and tear with the owner or service station. As IoT is becoming widespread, more and more sensors will become IoT enabled. We’ll look at this in more detail in the next section to give you a deeper understanding of possible application scenarios and affected sectors.
