Smart event-triggered communication for interconnected cyber-physical systems
As large, interconnected systems like smart grids and intelligent transportation networks grow, traditional control methods fall short. For his PhD research Koen Scheres addresses the challenge by developing robust event-triggered communication (ETC) mechanisms for cyber-physical systems. Key findings include a systematic approach to designing noise-resilient event-based controllers, a framework for ensuring stability through dissipation inequalities in time-based and continuous controllers, a distributed state estimation method that withstands measurement noise, and tools for designing effective local controllers in networked systems. These contributions support efficient, stable, and scalable control for complex, networked infrastructures.
As systems become (physically) larger and more interconnected, the typical classical control approach in which systems are regulated from one central unit becomes increasingly challenging. To keep up with the societal demands for smart interconnected systems -- such as intelligent transportation systems and smart power grids -- novel techniques to coordinate these systems and regulate the communication among subsystem is therefore paramount.
Design cyber-physical systems increasingly important
In many cases, the communication between (sub)systems is realized via a (wireless) packet-based network. The information that is exchanged via the network is often not only used to monitor and assess developments, but also to manipulate the physical world via digitally controlled actuators. The design of the resulting systems that rely on the interplay between the physical and digital domain, also referred to as cyber-physical systems, is extra challenging due to the inherent properties of the (wireless) networks that are used to exchange data. For example, the amount of data that can be transferred within a certain time span may be limited, the data may arrive with a delay or even not at all. In short, the design of cyber-physical systems is becoming increasingly important but remains far from trivial.
Reduce network utilization
Given the benefits and challenges described above, the design of controllers for cyber-physical systems, where the data is exchanged via shared (wireless) communication networks, has become a prominent topic within the control community. As the reliability of these networks often decreases with the amount of data that is exchanged, it is important to reduce the network utilization as much as possible, while still ensuring the desired system properties. To that end, event-triggered communication (ETC) schemes have been proposed in the literature. Contrary to classical digital control setups, these communication schemes exchange data aperiodically based on current output measurements. The rationale behind this method is that ETC schemes only utilize the network when actually needed to guarantee the desired system properties. Although a significant amount of research has focused on the stability of these ETC schemes, two highly relevant challenges have received little attention so far. The first is the presence of measurement noise, for which it is known that, in some cases, consecutive communication events may be generated arbitrarily close to each other in time, thereby rendering the scheme practically useless. The second is the ability to guarantee not just the stability of the system, but also its performance. The latter is especially important when specific properties are required of the closed-loop system. For example, in intelligent transportation systems such as vehicle platoons, we want to ensure so-called string-stability, formalized as a finite performance gain, to guarantee that consecutive vehicles do not need to break faster than their predecessor to avoid unnecessary traffic jams or even collisions.
Control and estimation
In his thesis presents a general framework for the design of event-triggering mechanisms for interconnected cyber-physical systems. In the first part, he mainly focuses on the design of event-triggering mechanisms that are robust to measurement noises, both in the context of control and in the context of estimation. In the second part, he emphasizes on the design of event-triggering mechanisms that guarantee desirable performance properties of the closed-loop system.
Title of PhD thesis: . Supervisors: Prof. Maurice Heemels, Dr. Romain Postoyan, and Dr. Victor Dolk..