ALSOS-ICS addresses this problem by providing an application-level self-organization model for control systems. More specifically, composition and reconfiguration are modeled as a control system. A software platform or middleware for distributed control systems can be used to build control applications that act as services for the composition and reconfiguration of other control applications. Therefore, ALSOS-ICS mainly introduces a practice rather than a new mechanism for self-organized control systems. This approach provides a higher integration, modularity, applicability and customization of various self-organization mechanisms in the context of distributed control systems.
The ALSOS-ICS model is outlined as follows:
- A discovery sensor senses for possible I/O bindings that can be configured between the control elements of the target control application.
- A decision controller selects I/O bindings that are added to or removed from the target control application based on criteria of this application.
- A reconfiguration actuator applies an I/O binding addition or removal between control elements of the target control application.
This model of control elements can be instantiated as a control application. Its goal is the binding management of a target control application. Compared to (i) the introduction of an external mechanism that may not be able to interoperate with the target control application or (ii) the integration of complex self-organization mechanisms within the target control application, this approach is more flexible, modular and customizable. It introduces a new type of control applications and control application-developers resulting in a split of concerns and a more in-focus development phase.
The ALSOS-ICS model can be instantiated in three granularity levels: (i) the system-level, (ii) the node-level and (iii) the element-level. In the system-level granularity, one or more ALSOS-ICS models are instantiated in a dedicated cyber-physical infrastructure different than the one of the target control application. In the node-level granularity, an ALSOS-ICS model controls the binding of the control elements located in a specific physical node. Finally, the element-level granularity introduces an ALSOS-ICS model for every control element of the target control application. The element-level provides the highest degree of decentralization and compositional autonomy.
ALSOS-ICS is instantiated at a node-level granularity in the Internet-scale Control System (iCS). iCS is a lightweight software platform for cyber-physical embedded systems written in the Java Micro Edition (JME) and developed at IBM Thomas J. Watson Research Center. iCS has been used in large-scale control applications of the Smart Power Grid such as energy demand-side management in the Olympic Peninsula Smart Grid Demonstration project and is currently the core communication technology in the Pacific Northwest Smart Grid Demonstration Project. Note that dynamic composition and reconfiguration of control between assets of the Smart Power Grid is one of the requirements of the Pacific Northwest Smart Grid Demonstration Project.
ALSOS-ICS has resulted in a conference publication. It was the main research focus of my internship at IBM Thomas J. Watson Research Center.
- Evangelos Pournaras, Mark Yao, Ron Ambrosio, Martijn Warnier, Organizational Control Reconfigurations for a Robust Smart Power Grid, book chapter in Internet of Things and Inter-cooperative Computational Technologies for Collective Intelligence, in the Springer Book Series ”Studies in Computational Intelligence”, Vol. 460, Chap. 8, pp. 189-206, January 2013. © Springer-Verlag
- Evangelos Pournaras, Mark Yao and Ron Ambrosio, Dynamic Composition and Reconfiguration of Internet-scale Control Systems, in the proceedings of the International Conference of Digital Ecosystems and Technologies-DEST 2011, pages 233-240, Daejeon, South Korea, June 2011. © IEEE [Presentation]