Current trends and future directions in event-based control

Michael Hertneck, David Meister, Frank Allgöwer

TL;DR

Event-based control (EBC) saves computational and network resources by updating feedback loops only when necessary, rather than on a rigid periodic schedule. This comprehensive review synthesizes current theoretical trends in EBC—such as unifying analysis frameworks, sampling behavior, and data-rate limitations—while charting out the most pressing open questions for future research.

What Problem Does This Paper Address?

Traditional control systems rely on Time-Triggered Control (TTC), where sensors sample and transmit data at fixed, periodic intervals. While mathematically straightforward, TTC is highly inefficient for Networked Control Systems (NCS) or battery-powered devices, as it wastes bandwidth and energy transmitting data even when the system is perfectly stable.

Event-based control solves this by only closing the feedback loop when a specific "triggering condition" is met (e.g., when the system state deviates beyond a certain threshold). The field is generally split into Event-Triggered Control (ETC), which continuously monitors the system to react to disturbances, and Self-Triggered Control (STC), which proactively calculates the next sampling time based on current data.

However, the rapid explosion of EBC research has led to a fragmented landscape of highly specific triggering rules and isolated mathematical proofs. The authors address the need to unify these theories, better understand the complex sampling behaviors EBC creates, and rigorously prove when EBC actually outperforms traditional periodic control.

Methods

As a review paper, the "methods" discussed are the state-of-the-art theoretical frameworks used to analyze EBC systems. The authors highlight a shift away from custom proofs for individual triggering rules toward unifying analysis frameworks. Key frameworks reviewed include:

  • Hybrid Systems Formalism: Modeling the EBC system as a hybrid system with continuous "flows" (between samples) and discrete "jumps" (at sampling instants).
  • Hybrid Small-Gain Theorems: Viewing the closed-loop system as an interconnection of two subsystems—one describing the plant/controller dynamics, and the other describing the sampling-induced error.
  • $L_p$ Norm Frameworks: Utilizing classical small-gain arguments based on $L_p$ norms to guarantee stability across different triggering rules.
  • Finite State Abstractions: A numerical method for analyzing sampling behavior by dividing the state space into conic regions and mapping the transitions between them to predict inter-event times.

Datasets

Not discussed in the paper. (This is a theoretical review paper, so no specific datasets are used for training or evaluation).

Key Insights and Ideas

  • The "Greedy" Nature of ETC: ETC algorithms typically act greedily, trying to maximize the time until the very next sample. Surprisingly, triggering a sample slightly earlier than strictly necessary can sometimes lead to longer average inter-event times over the long run.
  • Consistency over Optimality: Finding a mathematically "optimal" EBC scheme (balancing control cost vs. sampling cost) is notoriously difficult due to the hybrid nature of the math. A more pragmatic, emerging approach is designing for consistency—proving that an EBC scheme strictly outperforms the best periodic controller at the same average sampling rate.
  • Timing is Information: In rate-limited communication channels, EBC has a hidden superpower: the exact timing of a trigger carries implicit information about the system state. Because a packet was sent now rather than later, the receiver learns something about the state's magnitude, allowing EBC to stabilize systems using fewer transmitted bits than TTC.
  • Asymptotic Periodicity: While EBC sampling intervals can look chaotic initially, analytical tools for planar (2D) systems prove that they often converge to predictable, periodic patterns or constant intervals over time.

Key Results

Because this is a survey paper, it highlights benchmark theoretical results established in the literature rather than new experimental data:

  • Quantifiable Performance Advantage: For 1D stochastic systems, an optimal absolute threshold ETC scheme yields a performance advantage (cost reduction) of exactly $1/3$ compared to the best periodic controller operating at the same average sampling rate.
  • Zero-Bit Stabilization: In ideal rate-limited channels (no unknown delays), ETC can theoretically stabilize a linear system with an Information Transmission Rate (ITR) of zero after the initial setup, because the timing of the triggers conveys all the necessary state information.
  • Delay Penalties: Once unknown communication delays are introduced, the data-rate advantage of ETC over TTC shrinks. If delays are large enough, ETC may lose its transmission rate superiority entirely.

Limitations and Future Work

Limitations: * Current analytical tools for predicting exact sampling behavior are mostly restricted to simple setups (e.g., scalar or planar linear systems). * Numerical approaches (like finite state abstractions) suffer from the curse of dimensionality and scale poorly to complex, high-dimensional systems. * Many EBC designs assume centralized information, and their performance guarantees often vanish in decentralized or distributed setups.

Future Work: * Dissipativity Theory: Developing even broader, unified frameworks based on dissipativity theory or robust control (e.g., Integral Quadratic Constraints) to evaluate EBC schemes. * Standardized Benchmarks: Creating community benchmarks to fairly compare the performance-to-sampling-rate trade-offs of different EBC schemes against each other and against optimal periodic controllers. * New Domains: Applying EBC concepts to emerging fields that naturally fit the paradigm, such as event-based vision (neuromorphic cameras) and data-driven predictive maintenance.

Summary updated

Abstract

The defining characteristic of event-based control is that feedback loops are only closed when indicated by a triggering condition that takes recent information about the system into account. This stands in contrast to periodic control where the feedback loop is closed periodically. Benefits of event-based control arise when sampling comes at a cost, which occurs, e.g., for Networked Control Systems or in other setups with resource constraints. A rapidly growing number of publications deals with event-based control. Nevertheless, some fundamental questions about event-based control are still unsolved. In this article, we provide an overview of current research trends in event-based control. We focus on results that aim for a better understanding of effects that occur in feedback loops with event-based control. Based on this summary, we identify important open directions for future research.
BibTeX
@article{hertneck2026current,
  author = {Michael Hertneck and David Meister and Frank Allgöwer},
  title = {{Current trends and future directions in event-based control}},
  journal = {Eur. J. Control},
  year = {2026},
  doi = {10.1016/j.ejcon.2025.101387},
}
Figure