Industrial Internet: Towards Interoperability and Composability
communication protocols, data syntax and semantics to enable effective and high quality analytics. It is highly desirable to establish standards at the lower stack providing common data communication protocols and abstract data syntactic and semantic models across industrial sectors, while at the same time allowing specific syntactic and semantic models to be instantiated based on the common abstract models for each industrial sector according to their unique requirements.
3. The need for interoperability between the Information and the Application Domains to concisely represent analytic results.
4. The need for interoperability between the Application Domain and the industrial assets in the Control Domain so that commonly understood semantics can be relied upon when the high level applications interact with the industrial assets to provide feedback to their operations. When providing feedback to the industrial assets, composability is generally required in that the Application must have a level of expectation on how a specific asset will react to the requests it issues.
The most demanding challenges in interoperability lie in the Control Domain where functional components from various assets interact with each other, as outlined previously. How to meet these challenges is a key to the success of the Industrial Internet in the long term and yet requires substantial research and experimentation to achieve.
5. COMPARING INDUSTRIAL INTERNET SYSTEMS
In this section, we focus on two different approaches to Industrial Internet systems and point out composability issues. The first approach is the path most of us are embarked upon, but it illustrates the danger of presuming composability can be solved by standardization or brute force at a global level – requiring buy-in to a particular vendor framework or a particular set of standards. The second approach instead encourages an open market where composability is addressed locally within communities of interest rather than globally, encouraging engineering automation and specialization. We do not yet have all of the tools we need for the second approach, but it may be time to invest!
5.1 Short-Term Approach: Analytics in the Cloud
A common approach toward an Industrial Internet system, today, is to collect data( as much as possible) from every asset, store it“ in the cloud” for analysis, mine it to discover correlations and associations and then take the result and turn it into a new control or operating policy. This new control or policy can, for instance, optimize the life of the system, minimize downtime, improve asset power efficiency, etc. Such an approach is incremental over the kinds of things we have already been doing with assets. For instance, condition-based maintenance requires understanding the behavior of a component over its lifecycle. So that, given some sensor data, we can determine where it is in that lifecycle – nearly new or about time to replace, based on,
- 70- June 2016