Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Robust Event-Triggered Integrated Communication and Control with Graph Information Bottleneck Optimization

Ziqiong Wang, Xiaoxue Yu, Rongpeng Li, Zhifeng Zhao

TL;DR

This work tackles the challenge of efficient, consensus-driven communication in decentralized multi-agent reinforcement learning under partial observability. It introduces CDE-GIB, a framework that fuses a Graph Information Bottleneck regularizer with a variable-threshold event-triggered mechanism to compress and selectively transmit messages while preserving consensus quality. The approach jointly optimizes the communication graph and information flow, enabling concise yet sufficient representations for coordination, and it uses MAPPO as the learning backbone with centralized training and decentralized execution. Across simulations in the multi-agent particle environment, CDE-GIB outperforms state-of-the-art baselines in both communication efficiency and adaptability, demonstrating the practical value of information-theoretic graph compression in distributed control tasks.

Abstract

Integrated communication and control serves as a critical ingredient in Multi-Agent Reinforcement Learning. However, partial observability limitations will impair collaboration effectiveness, and a potential solution is to establish consensus through well-calibrated latent variables obtained from neighboring agents. Nevertheless, the rigid transmission of less informative content can still result in redundant information exchanges. Therefore, we propose a Consensus-Driven Event-Based Graph Information Bottleneck (CDE-GIB) method, which integrates the communication graph and information flow through a GIB regularizer to extract more concise message representations while avoiding the high computational complexity of inner-loop operations. To further minimize the communication volume required for establishing consensus during interactions, we also develop a variable-threshold event-triggering mechanism. By simultaneously considering historical data and current observations, this mechanism capably evaluates the importance of information to determine whether an event should be triggered. Experimental results demonstrate that our proposed method outperforms existing state-of-the-art methods in terms of both efficiency and adaptability.

Robust Event-Triggered Integrated Communication and Control with Graph Information Bottleneck Optimization

TL;DR

This work tackles the challenge of efficient, consensus-driven communication in decentralized multi-agent reinforcement learning under partial observability. It introduces CDE-GIB, a framework that fuses a Graph Information Bottleneck regularizer with a variable-threshold event-triggered mechanism to compress and selectively transmit messages while preserving consensus quality. The approach jointly optimizes the communication graph and information flow, enabling concise yet sufficient representations for coordination, and it uses MAPPO as the learning backbone with centralized training and decentralized execution. Across simulations in the multi-agent particle environment, CDE-GIB outperforms state-of-the-art baselines in both communication efficiency and adaptability, demonstrating the practical value of information-theoretic graph compression in distributed control tasks.

Abstract

Integrated communication and control serves as a critical ingredient in Multi-Agent Reinforcement Learning. However, partial observability limitations will impair collaboration effectiveness, and a potential solution is to establish consensus through well-calibrated latent variables obtained from neighboring agents. Nevertheless, the rigid transmission of less informative content can still result in redundant information exchanges. Therefore, we propose a Consensus-Driven Event-Based Graph Information Bottleneck (CDE-GIB) method, which integrates the communication graph and information flow through a GIB regularizer to extract more concise message representations while avoiding the high computational complexity of inner-loop operations. To further minimize the communication volume required for establishing consensus during interactions, we also develop a variable-threshold event-triggering mechanism. By simultaneously considering historical data and current observations, this mechanism capably evaluates the importance of information to determine whether an event should be triggered. Experimental results demonstrate that our proposed method outperforms existing state-of-the-art methods in terms of both efficiency and adaptability.

Paper Structure

This paper contains 12 sections, 4 theorems, 20 equations, 4 figures, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. System Model and Problem Formulation
  3. System Model
  4. Multi-Agent Proximal Policy Optimization
  5. Problem Formulation
  6. The Framework of CDE-GIB
  7. Variable Threshold-Event Triggered Mechanism
  8. Graph Information Bottleneck for the Communication Graph and Information Flow Optimization
  9. Simulation Settings and Results
  10. Simulation Settings
  11. Simulation Results
  12. Conclusions

Key Result

Lemma 1

For two random variables $X$ and $Y$,

Figures (4)

  • Figure 1: Illustration of MARL information control.
  • Figure 2: The overall framework of CDE-GIB.
  • Figure 3: Learning curves of consensus algorithms with and without GIB optimization.
  • Figure 4: Performance Comparison of consensus algorithms with and without ETM.

Theorems & Definitions (5)

  • Lemma 1: Nguyen, Wainright & Jordan’s bound GIB
  • Theorem 1
  • proof
  • Lemma 2: Ref. GMM
  • Corollary 1