A network architecture for IoT combines layers of technology -- including embedded sensors and core service layers -- that are critical to the success of an IoT deployment.
In today's world, organizations are still unraveling IoT and designing their IoT network architecture based on current integration points rather than choosing an out-of-the-box network structure.
A successful setup depends on the type of industry or market served by the IoT applications, said Arti Bedi Pullins, founder and CEO of Pundit Consultantz in Chicago, in an email. However, most IoT infrastructure plans involve two or three layers that ultimately relay data -- either in real time or in batches -- and analytical measurements based on the data the sensor or embedded system collects.
IoT gathers and analyzes real data from the physical world and translates it into workable uses based on industry; market; consumer; or even AI, algorithmic and programmatic needs, Bedi Pullins said.
The architecture combines cloud-based data centers; core services layers; connective layering, such as Ethernet, 5G and 4G; and embedded and sensory-based learnings, Bedi Pullins said. IoT network architecture includes a few simple layers:
Collection. This is what the device or sensor on the product side collects and measures.
Operational. These are the services and connector mechanics that will sit in the middle to connect and absorb all the calls, creating a gateway.
Distribution. This is the end layer and how the first two layers connect and deliver data and measurements in a meaningful and absorbable way.
Six IoT network architecture components
People will break up a network architecture framework for IoT into different areas, according to Tim Zimmerman, analyst at Gartner.
"Of course, the first item is going to be the IoT device itself, obviously," he said. "It could be a sensor. It could be an MRI machine in healthcare. It could be a lot of things."
The second component is communication -- how a device communicates its data -- Zimmerman said.
There are two basic enterprise network architectures that address IoT network communication infrastructure requirements for IT organizations: wide-area communication and cloud application or on-premises communication, according to a Gartner research note co-authored by Zimmerman. Wide-area communication is typically cellular-based, including low-power WAN (LPWAN). Cloud application or on-premises communication consists of many forms of local area wireless, including Wi-Fi and private cellular.
Depending on the business, the communication technology could be a cellular network, such as with autonomous cars, Zimmerman said. In smart cities, for example, sensors on lampposts use LPWAN technology to gather data and communicate with a central control system that makes automated decisions about when they should be turned on and for how long.
The third framework component is security, according to Zimmerman. Security technologies are necessary to protect IoT devices and platforms from breaches. Connected devices that have been in use for many years must communicate safely and securely with newer connected devices.
"[There are] devices that are now coming onto the network that, historically, we probably never would have thought about," Zimmerman said. "From an IT standpoint, this includes things such as your HVAC, your video camera, building automation. There are [a number of ways] that people actually try to secure these things."
The fourth IoT network architecture component is the gateway, he said. Gateways can house the application logic, store data and communicate with the internet for the things that are connected to it, according to Gartner.
Patrick Filkins, an analyst at IDC, said the primary function of an IoT network gateway is to perform protocol conversion.
"That means you may have a sensor that connects to the gateway using some sort of networking protocol, Wi-Fi, Zigbee, Bluetooth, even wired," Filkins said. "And it will take that local area connection and convert it to a much more efficient, longer wide area network backhaul connection that could have fiber optic. It could have cellular."
While all the components of the framework will exist, the complete solution -- the way you connect the dots -- is going to be different depending on the answers.
Tim ZimmermanAnalyst, Gartner
The fifth component is the IoT platform, which is an aggregation point for one or multiple different sites or products, Zimmerman said.
"This is where a lot of the data is collected. It tends to be where some of the upper-layer logic resides that may communicate to the gateway, but [it] also may provide direction to the applications," he said.
The final component is the application, which is kind of the user interface, according to Zimmerman. The application component uses collected data to enable users to monitor and control their cars or smart homes, for example.
Business requirements determine component arrangements
"So, there are different ways, depending on the [business] applications, that these things get stretched," Zimmerman said. "All the components exist, but where they reside may differ just a little bit depending on what the outcome needs to be. When I work with clients, it's important to understand what the business outcome is that they're trying to achieve and then look for all the components that may affect some of the decision-making capabilities within the framework."
In healthcare, consumer-based health data collected directly from physician-based sensors embedded within a stethoscope, glove or patient beds can interface with the operational side of the health system to deliver meaningful information, Bedi Pullins said.
However, the level of personally identifiable information and patient-level data security will be structured completely differently than for autonomous vehicles, she added.
Watch this short overview of IoT architecture and
Zimmerman said that before building or adding to existing systems there are certain questions an organization should answer regarding its proposed IoT network architecture. What kind of communication will be used? What is the distance that will be covered? Where will the sensor be located? How often will you communicate with the sensor? What kind of sensor is it? What kind of security does it have? Do you have a security policy?
"Because there are so many ways you can connect the dots, these are all the precursor types of questions that you really have to answer upfront," he said. "While all the components of the framework will exist, the complete solution -- the way you connect the dots -- is going to be different depending on the answers."