On the trade-off between event-based and periodic state estimation under bandwidth constraints

Dominik Baumann, Thomas B. Schön

Abstract

Event-based methods carefully select when to transmit information to enable high-performance control and estimation over resource-constrained communication networks. However, they come at a cost. For instance, event-based communication induces a higher computational load and increases the complexity of the scheduling problem. Thus, in some cases, allocating available slots to agents periodically in circular order may be superior. In this article, we discuss, for a specific example, when the additional complexity of event-based methods is beneficial. We evaluate our analysis in a synthetical example and on 20 simulated cart-pole systems.

Introduction

In networked control systems, such as drone swarms or automated factories, communication bandwidth is a critical resource. To maintain stability, sensors must transmit state updates to remote estimators. Traditionally, this is achieved through periodic scheduling (e.g., Round-Robin), where every agent receives a transmission slot in a fixed order. Recently, event-based methods have gained traction by promising to transmit data only when "necessary," such as when an estimation error exceeds a specific threshold.

However, event-based methods introduce a "cost of intelligence." They require a scheduler to prioritize agents, and the exchange of priority information itself consumes bandwidth. This research investigates a fundamental trade-off: is the benefit of prioritized scheduling worth the reduction in total available data transmission slots?

Methodology and System Architecture

The study focuses on Linear Time-Invariant (LTI) systems with process noise, comparing two primary strategies:

  1. Periodic (Round-Robin): A simple, predictable approach where every agent receives a slot in circular order. It features zero scheduling overhead, allowing all available bandwidth ($K_{per}$ slots) to be used for state data.
  2. Predictive Triggering: A "smart" event-based method that identifies systems with the highest estimation errors and prioritizes them. Because agents must first share their priority levels with the scheduler, this method can only afford fewer data slots ($K_{pred} < K_{per}$) per round.
System architecture for networked state estimation featuring a scheduler that manages transmissions across a wireless network.

The authors derived mathematical upper bounds for the expected quadratic estimation error for both schemes. Based on these bounds, they developed an Adaptive Scheme capable of switching between periodic and event-based communication in real-time as system dynamics evolve.

Experimental Results

The proposed framework was evaluated using synthetic LTI simulations and 20 simulated cart-pole systems within the MuJoCo physics engine.

Comparison of estimation error ||ê||² for periodic, predictive, and adaptive schemes. At k=100, a change in system dynamics causes periodic error to spike, while predictive and adaptive methods maintain stability.

Key Findings:

  • The Homogeneity Penalty: When all systems in a network are identical and stable, event-based communication is often suboptimal. The scheduling overhead provides no useful prioritization benefit, yet reduces the total number of successful state updates.
  • Heterogeneity and Localized Disturbances: Event-based methods excel when systems are diverse or face localized disturbances. In the cart-pole test, when the mass of 5 out of 20 carts was reduced (making them more volatile), predictive triggering significantly outperformed the periodic scheme by reallocating slots to the unstable carts.
  • Adaptive Success: The proposed adaptive scheme successfully identified shifts in system dynamics and switched to the optimal strategy, maintaining the lowest overall error across simulations.

Conclusion

The "intelligence" of event-based scheduling is most beneficial in heterogeneous environments where the gain from prioritization outweighs the loss of raw bandwidth. Conversely, simple periodic schedules remain superior for uniform, stable systems. Future work will explore average-case performance measures, the impact of non-reliable networks (packet loss), and the handling of large-scale disturbances that affect all agents simultaneously.

Summary updated

Abstract

Event-based methods carefully select when to transmit information to enable high-performance control and estimation over resource-constrained communication networks. However, they come at a cost. For instance, event-based communication induces a higher computational load and increases the complexity of the scheduling problem. Thus, in some cases, allocating available slots to agents periodically in circular order may be superior. In this article, we discuss, for a specific example, when the additional complexity of event-based methods is beneficial. We evaluate our analysis in a synthetical example and on 20 simulated cart-pole systems.
BibTeX
@inproceedings{baumann2023trade,
  author = {Dominik Baumann and Thomas B. Schön},
  title = {{On the trade-off between event-based and periodic state estimation under bandwidth constraints}},
  booktitle = {IFAC-PapersOnLine},
  year = {2023},
  doi = {10.1016/j.ifacol.2023.10.168},
}
Figure