Let’s get off the cloud! IIoT architectures for infrastructure environments.

With the revolution brought about by the new paradigms of IoT and Industry 4.0, there is no article that does not refer to Machine Learning, Big Data, Network Function Virtualization, Cloud Computing, Software Defined Networks, and a host of other concepts that, once put into practice, promise a more than promising future.

However, it is the task of each user to understand how to apply each idea in their work environment since, for each of these concepts, there is a very different implementation and use for each scenario. It is not the same to talk about Big Data for industrial environments, infrastructure, telecommunications, home automation, or large IT structures. In each of these environments, the idea is evidently the same, but its implementation is not.

In this specific post, we are going to discuss the concepts of Fog and Edge Computing, what they are, what differences they have, and how they are applied in a scenario where a set of geographically distributed facilities is managed.

Both concepts refer to control or information acquisition architectures, differing on what role local control elements have and what role a computing cloud has in the network.

Fog Computing.
In a Fog Computing architecture, the data that is acquired is sent directly to an intermediate infrastructure. This acts as a repository for other elements of the system to exploit this data and make action decisions, if necessary.

The idea is to create a large repository that is typically accessible via the Internet or private networks. Data exploitation systems – GeoSCADAs, historians, ERPs…- must have access to that contextualized data. These do not read a simple value from the repository, but a value with a timestamp and a node identifier.

In a Fog architecture, it is also possible for the infrastructure itself to have processing capacity, that is, to be able to make decisions based on the information received. However, today this functionality is not usually widely used in the world of facility management, but is more common in telecommunications or even the industrial world.

In summary, the field elements send the information to an intermediate infrastructure that acts as a gateway between them and the other elements of the system, adding security, user management, and contextualizing the data.

The technologies that perfectly fit this type of system today are those that use LPWAN communications – Low-Powered Wide Area Networks -, that is, SIGFOX and LoRA

Edge Computing
The concept of Edge computing is based on pre-processing the data locally. Basically, there is an intermediate device located in the facility itself that, when collecting the field information, analyzes and contextualizes it to decide:

• Whether an action should be taken immediately.
• What information should be sent to the higher systems, either using an intermediate platform or directly.

In this way, local control of the facility is maintained.

In infrastructure environments this intermediate element is typically an RTU.

To manage facilities or sites, it is not a matter of betting between one architecture or another, but understanding where each one fits.

At first, it seems logical to focus on Edge architectures – and it is -, since it covers essential needs due to the limited communications that are usually had with the facilities, either in bandwidth or connection stability.

It is indisputable that, compared to a Fog architecture, an RTU guarantees autonomous control and security regardless of the state of communications, in the same way that pre-processed data already reaches the higher systems, that is, relevant information for control, supervision, and monitoring. That is why, as far as the control of a facility is concerned, the Edge architecture is the appropriate one.

However, that does not mean that the Fog working method has no place in facility management, but its use must be well located.

A Fog architecture makes sense when the data that must be uploaded to the intermediate infrastructure does not require immediate action from the same device. Frequent use cases can be temperature, humidity, presence sensors… it is clear that depending on the value they report, one way or another must be acted, but it will not be the same sensor that activates the HVAC systems, to give an example. The person responsible for doing so will be, evidently, the RTU after the control center communicates it. In this way, the advantages provided by new technologies such as SIGFOX and LoRA – low consumption and great coverage – can be exploited, without endangering the control of the facility

In summary, a Fog architecture can complement a control by Edge Computing when the information that is collected does not require immediate action by the device.

If you want to know more about edge architectures, check out this post about our LK Remote solution.