Towards a performance-oriented perspective on event-triggered control

In many modern control systems, sampling utilizes a limited resource. Examples
include networked control systems, involving transmissions over a limited network
resource, or energy-constrained systems, where sampling depletes a limited energy
storage. It is thus relevant to improve sampling efficiency in control systems while
still achieving control objectives such as control performance. This is possible by
moving from periodic to aperiodic sampling techniques. While sampled-data control
traditionally treats aperiodicity as a disturbance, the advent of Event-Triggered
Control (ETC) marks a turning point in this paradigm. In ETC, a triggering scheme
determines sampling instants based on system information at runtime, closing the
loop only when needed. A core question is how to design triggering schemes and
controllers which together provide advantages over periodic control in terms of
resource and control objectives. In fact, examples have shown that these advantages
cannot be taken for granted. However, the key factors influencing if an ETC scheme
outperforms periodic control in terms of control performance and triggering rate are
still poorly understood.
In this thesis, we therefore develop rigorous concepts for comparing triggering
schemes and identify such key factors by revealing fundamental setups in which an
ETC scheme fails to deliver the desired advantages. Thereby, we unveil new insights
into the underlying mechanisms rendering ETC performant and sampling efficient.

• By (author) David Meister