What does “semantic interoperability by design” mean?

The basics of machines interoperability

When humans want to exchange information, both the sender and the receiver of information must share, on a common medium, conventions on the methods to exchange information (the protocol) and on the meaning of exchanged information, this latest part being called the language. The language is made of a common vocabular and a common grammar. The same happens for machine, except the fact that machines don’t have any ways for self correcting wrongly formed information (yet).

This point is expressed in the graphic on the left. It shows the needed steps for achieving interoperability, derived from the GWAC, and considered as the interoperability layers of the SGAM ( IEC SRD 63200. Starting from the component level (the physical medium), the next step is the communication ability (the protocol). Then the information layer follows, raising the need of common vocabulary and grammar. It hads as well two additional layers of highest interests, and their presence is explained below.

Most of past systems used to just rely on communication protocol interoperability – A “must” which however hides many shortfalls

The presence of a communication protocol between devices based on standards like MMS, Modbus, or DNP3, offers a “limited” garantee that two devices can exchange information. As such, they don’t garantee that concerned devices have a common “understanding” of the exchanged information. Comparing with human interoperability, two people having a phone may speak to each other (protocol level), but, unless they share a same language, they won’t understand each other (information level).

So up to now, it was the role of the system integrator to reach such semantic interoperability. This was happening through assessing the datasheets of all concerned devices, interprating their content (possible language errors), and adding specific treatments in hard or soft gateways to achieve the needed semantic transformation.

System integrators semantic added value (here modelled in red) is needed to ensure the device semantic match

By absence of semantic conventions, beside the additional need of gateways,
the integration process is time-consuming (depending on the number of data to process), is error prone and consequently needs in-depth testing.
It makes the whole system very rigid – any changes in data or devices would lead to re-enter such an heavy integration process

And if semantic by design were present

Due to semantic convention agreed at design stage, device interfaces are built to match each other

The value of semantic conventions

In case semantic conventions are present, devices have been designed to match each other. This means that at machine level vocabulary and grammar for data exchange, of the concerned set of devices, is predetermined, within a specific subset. This risk of misalignment only remains in possible misinterpretations of few words.

At the time of integration, the need of semantic alignment has disappeared : If one device sends the “word” active power” with a given format, the recipient will know that the value reflects “active power”, the the agreed format will avoid confusion in the unit associated to this value.

Systems based on standard semantic conventions are faster to build (lesser integration)
and are of higher quality (no “gateways”).
They are more agile and robust for supporting future evolutions, especially when relying on international standards

The main international semantic standards
for the Energy domain

Three main international standards cover the energy domain from a semantic perspective, i.e. providing machine level dictionaries to enable machines to exchange meaningful information :