Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Semantic Communication and Control Co-Design for Multi-Objective Distinct Dynamics

Abanoub M. Girgis, Hyowoon Seo, Mehdi Bennis

TL;DR

The work addresses distributed control of heterogeneous systems under limited wireless resources by learning semantic dynamics that couple latent-space linearization with logic-encoded control rules. It introduces a two-phase Semantic Logical Koopman Auto-encoder that separately captures shared semantic dynamics (Dynamic Semantic Koopman) and system-specific logic (Logic Semantic Koopman) to enable transfer across distinct dynamics. Simulations on cart-pole systems show substantial reductions in communication samples and major gains in state prediction and control performance, demonstrating scalable co-design across heterogeneous platforms. The approach offers a principled, data-driven path to task-focused communication-cost reduction without sacrificing control quality.

Abstract

This letter introduces a machine-learning approach to learning the semantic dynamics of correlated systems with different control rules and dynamics. By leveraging the Koopman operator in an autoencoder (AE) framework, the system's state evolution is linearized in the latent space using a dynamic semantic Koopman (DSK) model, capturing the baseline semantic dynamics. Signal temporal logic (STL) is incorporated through a logical semantic Koopman (LSK) model to encode system-specific control rules. These models form the proposed logical Koopman AE framework that reduces communication costs while improving state prediction accuracy and control performance, showing a 91.65% reduction in communication samples and significant performance gains in simulation.

Semantic Communication and Control Co-Design for Multi-Objective Distinct Dynamics

TL;DR

The work addresses distributed control of heterogeneous systems under limited wireless resources by learning semantic dynamics that couple latent-space linearization with logic-encoded control rules. It introduces a two-phase Semantic Logical Koopman Auto-encoder that separately captures shared semantic dynamics (Dynamic Semantic Koopman) and system-specific logic (Logic Semantic Koopman) to enable transfer across distinct dynamics. Simulations on cart-pole systems show substantial reductions in communication samples and major gains in state prediction and control performance, demonstrating scalable co-design across heterogeneous platforms. The approach offers a principled, data-driven path to task-focused communication-cost reduction without sacrificing control quality.

Abstract

This letter introduces a machine-learning approach to learning the semantic dynamics of correlated systems with different control rules and dynamics. By leveraging the Koopman operator in an autoencoder (AE) framework, the system's state evolution is linearized in the latent space using a dynamic semantic Koopman (DSK) model, capturing the baseline semantic dynamics. Signal temporal logic (STL) is incorporated through a logical semantic Koopman (LSK) model to encode system-specific control rules. These models form the proposed logical Koopman AE framework that reduces communication costs while improving state prediction accuracy and control performance, showing a 91.65% reduction in communication samples and significant performance gains in simulation.
Paper Structure (13 sections, 19 equations, 5 figures)

This paper contains 13 sections, 19 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. System Model
  4. Control Model
  5. Wireless Communication
  6. Background Principles
  7. Koopman Operator
  8. Signal Temporal Logic
  9. Semantic Logical Koopman Auto-encoder
  10. Dynamic Semantic Koopman Model
  11. Logic Semantic Koopman Model
  12. Simulation Results
  13. Conclusion

Figures (5)

  • Figure 1: Illustration of multi-objective distinct and distributed control systems.
  • Figure 2: Training procedure of the semantic logical Koopman for controlling two systems with two main phases.
  • Figure 3: State prediction and communication samples for each system of proposed and baselines with $\text{SNR} =15$ dB for (a) $d = 4$ and (b) $d = 2$.
  • Figure 4: Average score values for each system of the proposed and baselines with $\text{SNR} =15$ dB for (a) $d = 4$ and (b) $d = 2$.
  • Figure 5: State prediction and communication samples for each system of the proposed and baselines for (a) $\text{SNR} =15$ dB and (b) $\text{SNR} =5$ dB.