If your company is like many organizations, it's actively engaged in or considering launching one or more IoT initiatives. It has a goal, a strategy and a desired outcome -- whether to drive revenue, cut costs or optimize business processes. And it might have already selected technologies and suppliers.
You might be wondering why either of those matter -- two reasons. First, organizations with an IoT architecture are significantly more successful than those without. Successful companies -- those that rank in the top third of all companies when it comes to saving money, driving new revenue or improving business processes via IoT -- are 34% more likely to have an IoT architecture than less successful firms.
But that's not just any architecture, which brings us to the second reason it's important. Successful companies are more likely to adhere to an architecture that includes both a general-purpose IoT framework and a specific, customized version for the specific IoT project, whether that project is smart cities, industrial IoT or facilities IoT.
What is IoT architecture?
The best place to start is to define what an architecture is. Fundamentally, an architecture is a diagram or model that comprises two parts: the key technology components that make it up and the relationship between those components.
In other words, an architecture is more than a list of necessary technology components, but it does start with that list. It goes on to define how these components interact or engage with each other.
Some further definition is in order. By technology component, we mean a system, device or piece of software that delivers a specific technical capability. The cloud isn't a technology component, though it might be where a technology component is located -- e.g., a firewall in the cloud. Components can be hardware, software, or a mix of the two; and they can be defined at a high level (firewall) or a very granular level (link-layer packet filtering). Defining the components at the right level is part of the challenge of crafting an architecture. For example, in an IoT architecture, those components are likely to include connected devices, smart devices, sensors and actuators.
What are the components of an IoT architecture?
At a high level, the components involved in an IoT architecture include four key components. (See Figure 1 below).
- The applications and analytics component. This is the piece that processes and displays information collected via IoT. It includes analytics tools, AI and machine learning and visualization capabilities. Technologies for this component range from traditional analytics and visualization packages, such as R, IBM SPSS and SAS, to specialized IoT tools and dashboards from cloud providers, such as Amazon, Google, Microsoft, Oracle and IBM, as well as application suite vendors, including SAP and Salesforce.
- The integration component. This is the component that ensures that the applications, tools, security and infrastructure integrate effectively with existing companywide ERP and other management systems. Providers include the aforementioned software and cloud players, as well as a range of open source and middleware providers, such as Oracle Fusion Middleware, LinkSmart, Apache Kafka and DynThings Open Source IoT Platform.
- The security and management component. IoT security includes securing the physical components of the system via firmware and embedded security providers, such as Azure Sphere, LynxOS, Mocana and Spartan. Traditional security vendors, such as Forescout, Symantec and Trend Micro, also offer packages that focus specifically on securing IoT.
- The infrastructure component. This includes physical devices -- IoT sensors, which capture information, and actuators, which control the environment. It also includes the network on which the sensors or actuators reside. Typically, though not always, this is a wireless network, such as Wi-Fi, 4G or 5G.
The relationship between the components is the second part of the architecture. By relationship between, we mean how components communicate with each other, and what sorts of information are exchanged. This can include data flow, metadata flow, control information or no information at all. Software components often communicate via APIs, and network layer components typically communicate via network protocols.
Speaking of layers, architects often think in terms of component layers, such as network layer, perception layer, processing layer, physical layer, gateway layer, platform layer, device layer, business layer, security layer, sensor layer and so on.
The concept of a layer is that it comprises a set of capabilities that communicate with one another, but for the purpose of other components can be treated as a single entity, with a single transparent entity.
In IoT architecture, the application layer need not know what type of physical network carries the data. All the network devices comprise the network layer that transports traffic as needed by the applications.
What are the 6 layers of IoT architecture?
We define the six layers of IoT architecture as described below. Note that in some cases, layers are made up of sublayers. This a common characteristic in complex architectures, such as that of IoT. (See Figure 2 below).
- Physical/device layer. This comprises the sensors, actuators and other smart devices and connected devices that comprise the physical layer and device layer. These smart devices either capture data (sensors), take action (actuators) or sometimes both.
- Network layer. This comprises the network devices and communications types and protocols (5G, Wi-Fi, Bluetooth, etc.). Although many IoT architectures rely on general-purpose network layers, there is an increasing trend to move to dedicated IoT-specific networks.
- Data/database layer. This also includes the database platform layer. There are a range of databases used for IoT architectures, and many organizations spend a fair amount of time selecting and architecting the right IoT databases.
Together, the physical layer/device layer, network layer and data/database layers comprise the infrastructure component discussed above.
- Analytics/visualization layer. This layer comprises the analytics layer, visualization layer and perception layer. In essence, this layer's focus is on analyzing the data provided by IoT and providing it to users and applications to make sense of.
- Application/integration layer. This is the layer of applications and platforms that integrate together to deliver the functionality from the IoT infrastructure to the business. In other words, the application layer, platform layer and integration layer are what provide the business value from the IoT infrastructure. The processing layer and business layer are all part of the larger application/integration layer.
- Security and management layer. As the name implies, this layer encompasses both the security layer and the management layer. Strictly speaking, this is not a layer as it has connections with all the other layers to provide security and management. But it's an important component that's worth considering at every layer.
These layers go from bottom to top in a fashion similar to the Open Systems Interconnection model, which was the original source of the layering concept. (See Figure 3 below).
Conclusion and recommendations
Enterprise IT, OT and IoT technology professionals should develop IoT projects and initiatives based on a consistent architecture. That doesn't mean using the exact same tools and technologies for every project, but rather making sure every component is properly instantiated for the specific project, and that technology professionals have thought about all the layers, including the network layer, perception layer, processing layer, physical layer, gateway layer, platform layer, device layer, business layer, security layer and sensor layer.