Industrial Internet: Towards Interoperability and Composability
them to interconnect to and interact with each other directly to enable autonomous
collaboration among themselves to solve problems locally. As a result, the existing point-to-point
interconnect patterns will give way to a meshed many-to-many ad hoc interconnectivity. Given
the large variations in the types of the CPSs expected to be involved in such interactions in a
deployment, the interoperability problem multiplies. This multitude of interactions among the
industrial assets raises a unique quantitative challenge in interoperability that we expect to face
in the Industrial Internet. A new class of cross-industrial sector connectivity frameworks such as
Data Distribution System (DDS)31 and OPC-UA32 are expected to be seen increasingly deployed
not only providing the needed interoperability in communication, connectivity and data
modeling but also scalability, performance and security as well.
Moreover, consider the many unique characteristics that are present in industrial systems,
including:
produce physical effects
strong safety requirements
often mission-critical
subject to dynamic conditions and operate in environments that are not always
foreseeable at the design time
long lifecycle, being very costly or even impractical to replace on demand
advancement towards autonomy, increasingly capable of learning and making decisions
based on conditions without intervention
All of these pose a qualitative challenge in interoperability in how to ensure the correct outcomes
in these interacting systems under changing conditions and dynamic environments. This
challenge calls for a higher level of interoperability: composability, as highlighted at the
introduction of this article.
Adding to these quantitative and qualitative interoperability challenges is the dynamism
of the system within itself concerning how to sustain the required level of interoperability as the
interacting components evolve over time in their respective lifecycles and as new components
enter into the system. To address these issues, it may be beneficial to reference the IT/Cloud
Computing world where the Service Oriented Architect ure (SOA) paradigm has matured and is
widely adopted in software design. Within this design paradigm, self-contained units of software
capability are encapsulated as services and exposed through service interfaces, often in the form
of APIs (Application Program Interfaces) that can be invoked remotely through a network. This
allows software capabilities to be loosely coupled (allowing independent evolution, among other
things) and large software capabilities can be assembled from a set of cooperating services
31
http://portals.omg.org/dds/
32
https://opcfoundation.org/about/opc-technologies/opc-ua/
IIC Journal of Innovation
- 65 -