The IIoT Architecture Design

The IIoT Architecture: How to Tap Into Its Full Potential?

The hype surrounding IoT technology and the sheer number of solutions—from highly customized, through do-it-yourself, to plug-and-play options—can be overwhelming. Rapidly diversifying offerings may not always readily engage with the question of how your infrastructure can support IoT. The basic IIoT architecture supporting IoT systems has four key interlinked constituents. In their togetherness, they form the IIoT process. We can also see the IIoT architecture as a four-stage process whereby insight-capable data moves from different networked things to traditional IT systems or to production where it generates value and business-relevant knowledge.

Setting up a solid framework for the Industrial Internet of Things (IIoT) is at the heart of a long-term, robust environment for IoT project development in industrial settings. A process view on IIoT acknowledges that this involves much more than simply having multiple industrial IoT devices connected to the internet. And a robust infrastructure is a decisive constituent of that process. 

Devices at the IoT edge 

These are your groups of networked things residing in the far end of an IoT network. These are located as close to the data source as possible. Typically, these are wireless sensors and actuators. We have a processing unit or a small computing device together with a group of observing endpoints. Devices at the IoT edge can be anything from legacy devices within a brownfield environment, sensors, robotic cameras, microphones, to all sorts of meters and monitors. 

What happens at the farthest edge of the IoT network? Sensors collect data, both from the environment or from the objects they are measuring. Then they convert it into actionable data that people can use. The actuators regulate the processes taking place within the measured environment. They adjust the physical conditions within which the data is generated. For example, an actuator may open or close a valve, or move a robotic arm as part of an automated assembly process. 

The gateway

The internet gateway is where sensor data is aggregated and converted into digital streams for future processing. After receiving the aggregated and digitized data, the gateway routes it via the internet so that it can be further processed before it gets to the cloud. Gateways are data acquisition systems that are still part of the edge. Depending on the taxonomy, they would still count as edge devices. They are still located close to the sensors and actuators, and pre-process the data at the edge. Gateway devices need to be portable, flexible, and able to withstand various environmental conditions.

Why do we need gateways? The data streams coming from the sensors accumulate massive amounts of constantly changing data in a minimum of time. Data is streamed continuously, creating enormous data flows that are simply too much to transfer directly to the cloud. Things get even more overwhelming when you have a multitude of sensors streaming data into an IoT system. Also, if the sensor data is in an analog form, it has to be converted into digital data to allow for future processing. This conversion takes place at the gateway, too. 

Apart from this basic gateway functionality, some gateways have analytics and data management services as well as some built-in security tools. Intelligent gateways make it possible to analyze the incoming data streams in real time. Analytics performed at the gateway are not as fast and immediate but at this stage, you have more compute power. 

Data management layers

You cannot make full use of advanced analytics and artificial intelligence without high-quality, high-volume data. Data processing can take place even at the sensor level, and this is great news if you are in immediate need of information. In this regard, edge computing delivers the fastest responses as data is pre-processed at the edge of the network, right at the sensors. 

Once the sensor data has been collected, converted into digital data, and aggregated, it is ready to undergo additional processing via edge IT systems. Edge IT systems can be on-premise or remote, but most commonly these reside close to the sensors. This is where you take your digitized, aggregated data to perform analytics on it. At this stage, you have data that makes sense. You have machine learning and data visualization. Once you have gleaned actionable insights from the data, you only need to move forward with the insights instead of passing on the entire information collected thus far. This additional processing reduces the data load going towards the data centers or to the cloud. In this way, you alleviate storage, security, and downtime concerns. 

Cloud (or data center)

But edge devices can only take a limited amount of pre-processing. While the effort should be to get as close to the edge as practically feasible to minimize the use of local computational resources, you will need to turn to the cloud for deeper and more comprehensive processing. At this stage, you may want to additionally decide whether you have to foreground the speed and immediacy that come with edge computing or the deeper insight that comes with cloud computing when processing that data. Comprehensive processing takes place in cloud systems. There, you can combine data from disparate sources and generate insights not immediately available at the edge.

Image 1. A basic overview of the IIoT architecture design

Here high-quality data is analyzed, managed, and stored on cloud, on-premise data centers, or hybrid systems. This is where you perform in-depth processing. You have enough power to extensively analyze and manage the data in a secure environment. Whereas it takes longer to get results, at this stage you can carry out truly comprehensive analyses. You combine insights from different sources to generate new knowledge.

A recent phenomenon: IIoT platforms 

Today, industrial IoT platforms (IIoT platforms) are able to orchestrate, monitor, and control processes across the entire value chain. Starting from the edge devices (gateways) and sometimes, at the very level of the sensors (Things), they manage the device data, take care of the analytics, data visualization, and AI tasks, all the way to the cloud (and back).

Computing and analytics at the edge have gained popularity over the past few years, mostly because they allow for speedy and immediate data management. A recent forecast analysis projects that by 2022, about 75% of all processing and data collection will take place outside the traditional data center. According to Gartner, by 2022, there will be more IIoT analytics performed at the edge than analytics on the cloud. 

IIoT platforms can intervene even here to take over a variety of functions around edge computing.

Once the data is extracted and pre-aggregated, you need to have the tools and processes in place to transform the data, employ advanced analytics, or train machine learning models. The data science functionalities of IIoT platforms can simplify these tasks.

But IoT data extraction, analytics, and device management are not all there is.

The ideal IIoT-enabling platform supplements its device management environment with app development capabilities that can reside in the cloud or on-premises. You develop IoT apps and build your own machine learning models right on the platform, programming on the IoT device and receiving feedback in real-time. Using container technologies, rolling out your industrial IoT applications on multiple devices is easy, as is the deployment of your new app versions.

Let’s chat about this and see how you can make the most out of your IIoT architecture.

Record Evolution Logo

About Record Evolution

We are a Data Science & IoT team based in Frankfurt am Main, committed to helping companies of all sizes to innovate at scale. So we’ve built an easy-to-use development platform enabling everyone to benefit from the powers of IoT and AI.