ABSTRACT
This chapter describes a hybrid event-based state estimator allowing both stochastic and deterministic representations of noise and estimation results. It presents an overview and outlook on state estimation approaches that do not require periodic measurement samples but instead were designed for exploiting a-periodic measurement samples. The estimator exploits the event-triggered sensor measurements to compute time-periodic estimation results, which are possibly used by a time-periodic controller for computing stabilizing control actions. Event-based sampling is an a-periodic sampling strategy where events are not triggered periodically in time but at instants of predefined events. The challenge in event-based state estimation is an unknown time horizon until the next event occurs, if it even occurs at all. The transmission rate can significantly be reduced if an event-based strategy is employed for sampling sensor data. In networked systems, high measurement frequencies may rapidly exhaust communication bandwidth and power resources when sensor data must be transmitted periodically to the state estimator.
