Modern IoT systems are often built by composing subsystems. Additionally, these systems may need to interact horizontally with other systems. This composition creates a new level of challenges for interoperability. Semantic interoperability is a collection of technologies that enable computer systems to interact unambiguously. It’s an important way to drive down the cost of integrating subsystems, but looking further into the future, it is also a core requirement for creating autonomous operations in IoT. Further, semantic interoperability greatly enhances the ability to build systems of independent IoT systems. Here, we explore the value of understanding semantic interoperability and the opportunities this creates.
The importance of the role of semantic interoperability in IoT
Semantic interoperability includes the ability to establish a shared meaning of the data exchanged, as well as the ability to similarly interpret communication interfaces. Shared meaning here means that two different computer systems, for example, not only can communicate data in the basic sense (such as an integer with value 42), but also attach unambiguous meaning to the data. For example, radiator three’s temperature in the conference room on level five is currently 42 Celsius.
As we build large IoT systems we are faced with several challenges of scale. Among them is the challenge of being able to make equipment and subsystems of different vendors interoperable and, over different time periods, work together and as intended. As it stands, the Internet protocol suite (IP, TCP/UDP, HTTP/CoAP, and the like) has done well in solving issues at the network level. However, these subsystems typically come with their own information model, data model, syntactic flavor, and concepts, which poses a problem at a higher level in integrating subsystems from multiple vendors.
And here lies the core semantic interoperability problem: How can the inherent meaning of a piece of data be preserved across different domains, without needing a human understand and ensure correct translation?
The lower the level of semantic interoperability between the subsystems, the more time consuming, costly, and error prone the effort will be to integrate and maintain these systems. Large parts of the effort will be manual work, which is hard to automate thanks to the required human interpretation of the meaning of the data involved.
The ideal scenario here? That these subsystems are able to solve this interoperability problem automatically.
Semantic interoperability can today be enabled by declarative models and logic statements (semantic models) encoded in a formal vocabulary of some sort. The fundamental idea is that by providing these structured semantic models about a subsystem, other subsystems can with the same mechanisms get an unambiguous understanding of the subsystem. This unambiguous understanding is the cornerstone for other subsystems to confidently interact with (in other words, understand information from, as well send commands to) the given subsystem to achieve some desired effect.
It's important to note that interoperability is beyond data exchange formats or even explicit translation of information models between a producer and a consumer. It’s about the mechanisms to enable this to happen automatically, without specific programming. There should be no need for an integrator to review thick manuals in order to understand what is really meant with a particular piece of data. It should be fully machine process able.
Today, industry standards exist that greatly improve interoperability with significantly reduced effort. They do so by standardizing vocabularies and concepts. However, such standards hold primarily within an industry or technical vertical for subsystems within a system. Providing horizontal interoperability—where it is unsure which other systems may interact with a given system—is a core challenge that semantic interoperability tries to address. This is not an imaginary requirement, but is the norm in the IoT world, where new applications will put together systems developed independently and for their own purpose.
Figure 1 shows interconnected things in a smart factory that enable the high-level interoperability of what a piece of data means.
Interestingly, an IEC report says how industry standards haven't really helped interoperability by themselves, even within an industry. There are many reasons for this, including that different domains come with different vocabularies and models, but also the fact that wider interoperability on data models becomes a substantial problem when the complexity of, and number of subsystems increase.
Semantic Interoperability in the manufacturing industry
The manufacturing industry is an area where interoperability is becoming crucially important, particularly as industry is moving towards Industry 4.0, with its emphasis on flexible production lines and reduced integration costs to introduce new production machines.
Let’s assume a manufacturing floor has several pieces of equipment and a new drilling machine is to be integrated into the process. The immediate problem is understanding the controls of the device as well as understanding the data it produces. In particular, the information model from the device needs to be understood by the floor management application, as it enables and supports better decision making, such as failure prediction. It also helps to avoid repeated programming efforts for each new device, as a semantic layer can be introduced that shields the algorithms from the specific information model of the device. The device itself may have its own vocabulary and then there’s the problem of mapping them. Often, the concepts do not map one-one and therefore, there is a significant transformation problem. These are the most basic challenges that semantic interoperability needs to deal with.
Semantic interoperability and smart city applications
A smart city consists of a range of systems such as transportation of different modes, buildings, utilities of different kinds, sewerage, water, waste management, and so forth. In order to meet future needs such as the sustainability, livability, and convenience of cities, the operations of each of these systems need to be optimized—and IoT plays a major role here.
Additionally, these systems need to be horizontally integrated to provide new applications, often through analyzing their data across systems. However, since these systems have their own information model and semantics, and also because each of them evolves independently over time, it becomes tricky to create new applications that can understand the information in these different systems and continue to work as the individual systems evolve.
One way to solve this is to ensure the platforms export information with semantic annotations. This significantly eases the burden of integrating information from these different sources and understanding them in a consistent way, which accelerates the development of new value-added applications.
Purposeful contributions and relevant standards for IoT
At Ericsson, we believe strongly in interoperability as a key enabler for unlocking the real value of IoT. In 2016, we hosted the Internet Engineering Task Force (IETF) Internet Architecture Board (IAB) workshop on Semantic Interoperability for IoT. The following year, we organized the first Internet Research Task Force (IRTF) Thing-to-Thing Research Group (T2TRG) Workshop on IoT Semantic/Hypermedia Interoperability (WISHI). The work has continued in the WISHI activity of the T2TRG. As a highlight, there have been several interoperability topics, such as LwM2M, IPSO, CoRAL, Web of Things (WoT) and One Data Model (OneDM), covered during the IETF hackathons that have followed.
One of the early organizations to address semantic interoperability challenges for IoT devices was the IPSO Alliance. We were a founding member of the IPSO Alliance, co-chairing the group responsible for defining the first IPSO object definitions for a set of common sensors and actuators. The IPSO Smart Objects are general in nature rather than specific to any industry vertical or particular application. This means that variety of use cases across multiple domain are addressed.
Later on, IPSO merged with the OMA SpecWorks and today, it continues as an OMA SpecWorks working group. We at Ericsson have continued to invest and engage with the IPSO work in OMA SpecWorks, driving the evolution of and promoting new, well-specified IPSO data models.
The W3C has defined various ontologies and other semantic interoperability enablers for the "big web" and more recently, has also addressed IoT use cases in the Web of Things (WoT) interest and working groups. The WoT Thing Description (TD) format enables us to describe interaction capabilities of various connected devices and things in a standard way, as well as use external vocabularies to give further semantics for the capabilities.
While the IETF has traditionally focused on network and transfer protocols, many activities have also addressed the interoperability of data. Sensor Measurement Lists (SenML) is an IETF standard for simple data model of IoT sensor and actuator data where we have been heavily engaged. One of the mechanisms for SenML is to provide interoperable hints of the semantics of the data using SenML units; standardized identifiers of the engineering units of the data. The SenML units have been adopted by various Standards Developing Organizations (SDOs) and other organizations for common representation of the unit identifiers.
Another Ericsson-driven activity for IoT in IETF is the Constrained RESTful Application Language (CoRAL), which defines a data and interaction model for IoT, and enables extensible semantics descriptions for the hypermedia controls through link relations. CoRAL is currently under development, but intends to further enrich the expressiveness of IoT data and interactions to improve interoperability of large systems.
For us at Ericsson, we have been raising awareness for some time that semantic interoperability is a key technology to enable large scale IoT deployments. In particular, it is an enabler for autonomous intelligent systems built on top of IoT. We have also seen that the interoperability problems are quite challenging, and that there is a need for technologies to solve several of the challenges that appear as a result. Efforts to solve semantic interoperability continue to evolve, and lately, a new initiative has been started that focuses on cross-ecosystem interoperability.