What is IoT Architecture?
The IoT architecture refers to the overall design and organization of connected devices, communication protocols, data processing, and storage technologies that make up the Internet of Things.
The architecture of an IoT system is to be considered in four stages. These stages can be listed as follows:
- IoT Devices (Sensors/Actuators)
- Edge IT
- Cloud/Data Center
In the first stage of an IoT architecture, we have IoT devices. These devices can be sensors (wired/wireless), actuators, etc.
In stage 2, we have data collection/aggregation systems. It may also include the ‘analog to digital’ data conversion step.
In stage 3, we have edge IT systems that may perform some sort of preprocessing on the data before the data is sent to the cloud/data center.
In the last stage 4, the data is stored, processed, and analyzed.
It is obvious that in the first stage, the persons involved will be professionals from operations technology. They are usually called OT people. It is also true with stage 2.
In stage 3 and stage 4, we have people from the IT domain. The edge IT processing may be located near the data center or it may be at a remote location.
Table of Content
5 Stages of IoT Architecture
Let’s describe the IoT (Internet of Things) architecture in detail.
Stage 1: IoT Devices
The data is collected by the sensors from the object under consideration. The physical conditions of the object or the surroundings are then modified or controlled by the actuators as per the decisions taken by processing the data.
Examples of sensing/actuating devices can be anything like industrial devices, camera systems, air quality sensors, water level indicators, accelerometers, pulse rate sensors, heart rate monitors, etc.
The list of IoT devices is continuously increasing with the expansion and focus on IoT systems. An IoT architecture, in some cases, may involve some sort of data processing at every stage. The amount and kind of processing that can be done at the IoT device level are limited by the available processing power on the IoT device. Most the IoT devices have got low power due to their design considerations.
Depending on the situation and the requirement, the data may need to be processed at the device level itself. For example, robotic surgery, car accident, etc. In such cases, we cannot have time to send the data to the cloud and wait for the response. So, the data needs to be processed there itself – at the device level. In other cases, the data may be sent to the cloud for storage and further processing.
Data is at the core of IoT system architecture. When processing the data, you have to decide whether to process it onsite or send it to the cloud.
Stage 2: Internet Gateway
Before the data is received by the gateway, it needs some sort of pre- processing. It is like aggregating all the data and converting the an- alog data to digital. There are systems called Data Acquisition Sys- tems (DAS) that perform the tasks of data aggregation and necessary conversion. The Internet gateway gets the data in the aggregated and digitised form. It is then routed via Wi-Fi, LANs, or the Internet, to the next stage 3 for further processing.
The system at stage 2, usually resides near the site, i.e. to the sensors and actuators. For example, a water pump may have some IoT devices that send data to the data aggregation/digitization system. This system might be attached to the pump physically. The received data might be then processed by the nearby gateway or server and forwarded it to the systems in the next stages 3/4.
The basic gateway functionality can be extended by adding other capabilities to it making it an intelligent gateway. This gateway would then be able to perform malware protection, analytics, data management services, etc. By using such systems, we can analyze in real-time of the data streams.
Gateways are external devices to the data center. They are like edge devices. Here the location and geography does matter. As in the pump example we have taken above, suppose there are 100 pump units and we want to process the data onsite. As we have instant data available at the pump level, we can aggregate it and create a plant-wide view. At the same time, we can send this data to the cloud/data center for a larger companywide view.
Because of DAS and gateways, we may have a wide variety of IOT devices from the factory site to the mobile stations, so the designing of these systems is generally kept portable, easy to deploy and maintain and at the same time rugged enough to sustain the temperature variations, dust, humidity, vibration, etc.
Stage 3: Edge IT
From stage 2, now the data enters stage 3. There may be, however, some data processing needed depending on the kind of application and domain before it is sent to the cloud/data center. This is the point where the edge IT systems get introduced. They further process the data and perform more analysis. These systems (edge IT) may be located anywhere, may be onsite, or in remote offices. Generally, they remain near the facility where the sensors are placed.
As the amount of data being generated from the IoT systems is very large, it will be unwise to send the whole of the data to the cloud. This is so because it will consume a lot of network bandwidth and other resources on the cloud. It is thus desirable to have some sort of preprocessing on the data at this stage so that only the needed data should be sent to the cloud for further processing. It will thus reduce the amount of data to be sent to the cloud, reducing the network bandwidth usage as well as the other resources on the cloud.
Security is another concern. If we have to send the whole data to the cloud, then the kind of security needed to protect the data will be very complex. So, it is always a good practice to have some powerful systems at the edge level to do some processing on the data, perform some sort of analysis and send only the data that needs further processing or which is required to be shared with others.
For example, as in the pump case, we can collect, convert and analyze data (e.g. vibration data) and send only the results/projections such as the device getting failed or in need of maintenance to the higher level.
Let’s consider another scenario. Suppose we have the machine maintenance data available with us. If we apply some sort of machine learning algorithm at this edge level and identify that some abnormalities are present that predict the approaching maintenance problems. Now this insight calls for immediate attention.
Here, we can use the visualisation systems to present that insight in a more meaningful way using graphs and dashboards which are easy to understand. This will help us in taking immediate action and prevent any further damage.
Thus, having an edge IT system help us in many ways.
Stage 4: Cloud/Data Center
The data is sent from gateway/edge IT to the Cloud/Data Center for further storage and processing. Here, we apply the various algorithms and models for further processing of the received data. At the cloud level, we have more powerful systems that can store the data at a large scale as well as perform analysis on the large data.
We can thus obtain a deeper insight into the given data and are in a position to make better decisions. Further, we can manage the data effectively and provide security to the data.
It may take some time to get the insight from this stage; however, the results obtained have more meaningful information and deeper insight.