REFLEXES: A Co-Designed Architecture for In-Network Control
Recent technological developments in sensing, communication, control and computation have fostered an emerging class of complex applications, called Cyber-Physical Systems (CPS). In these networks, a process is no longer bound to a specific independent device, but coordinated between several interdependent network nodes; and therefore distributed via a communication network. CPS are making inroads into an ever increasing number of application domains such as industrial automation (Industry 4.0), energy, transport, and health care systems. The deployment of CPS promises considerable benefits, including increased flexibility, efficiency, and adaptability.
However, the traditional separate and layered design of communication and control architecture prevents the ubiquitous adoption of CPS as the communication-induced unreliability and latency compromises the quality-of-control and may even lead to dangerous behavior. The core concept of the project has an analogy in nature: Within the human body reflexes are fast pre-defined simple action patterns, which are located in the spinal cord, i.e. close (in the sense of low transmission latency) to the sensors and actuators. Higher-level planning and control mechanisms are located farther away in the brain.
Therefore, the goal of this project is to develop a novel co-designed architecture for communication and control to facilitate the best possible performance of CPS given the available communication and computation resources. We introduce the novel paradigm of in-network control, which pushes control functionalities as close as possible to the process to be controlled exploiting the computational power of active network components - even if limited. We will move away from the ‘traditional’ layered communication architecture and deploy in-network processing of simple control commands bypassing the classical protocol stack. In short, the general objective of this project is to reduce the communication distance of control messages as much as possible (horizontally, in the number of hops, and vertically, in the time for processing) and to increase reliability by introducing deterministic processing of such messages. As a result unnecessary time delays and unreliability within the closed control loop are avoided leading to the best possible quality-of- control. The key methodical contributions of this project are i) a novel approach to distribute the control on a given communication and computation infrastructure including appropriate analysis tools for performance evaluation, ii) a novel software framework for in-network processing with real-time constraints by co-designing the execution of control and communication procedures in a single and fundamentally new methodology.