Research Analyzer
← Back ICRA 2026

Bandwidth-Efficient Multi-Agent Communication through Information Bottleneck and Vector Quantization

Ahmad Farooq, Kamran Iqbal

PDF

AI summary

Key figure (auto-extracted from paper)
Combining information bottleneck theory with vector quantization enables selective multi-agent communication that boosts coordination performance by 181.8% while cutting bandwidth usage by 41.4%.
Multi-Agent Reinforcement Learning Bandwidth-Efficient Communication Information Bottleneck Vector Quantization Gated Communication Robotics

Problem

Real-world multi-agent systems face severe communication constraints like limited bandwidth and energy, yet existing learned communication protocols lack principled ways to control usage while maintaining coordination effectiveness.

Approach

The framework uses information bottleneck theory to compress messages and vector quantization to discretize them, paired with a learned gating mechanism that dynamically decides when agents should communicate based on environmental context.

Key results

  • 181.8% performance improvement over no-communication baselines
  • 41.4% reduction in bandwidth usage
  • Dominance across the success-bandwidth Pareto frontier
  • Dual constraint enforcement mechanisms for flexible deployment

Why it matters

Enables reliable, scalable coordination for bandwidth-constrained multi-agent deployments such as robotic swarms, autonomous vehicle fleets, and distributed sensor networks.

Abstract

Multi-agent reinforcement learning systems de- ployed in real-world robotics applications face severe com- munication constraints that significantly impact coordination effectiveness. We present a framework that combines infor- mation bottleneck theory with vector quantization to enable selective, bandwidth-efficient communication in multi-agent environments. Our approach learns to compress and discretize communication messages while preserving task-critical infor- mation through principled information-theoretic optimization. We introduce a gated communication mechanism that dynam- ically determines when communication is necessary based on environmental context and agent states. Experimental evalua- tion on challenging coordination tasks demonstrates that our method achieves 181.8% performance improvement over no- communication baselines while reducing bandwidth usage by 41.4%. Pareto frontier analysis shows dominance across the entire success-bandwidth spectrum, with an area under the curve of 0.198 vs 0.142 for next-best methods. Our approach significantly outperforms existing communication strategies and establishes a theoretically grounded framework for deploying multi-agent systems in bandwidth-constrained environments such as robotic swarms, autonomous vehicle fleets, and dis- tributed sensor networks.

Index terms

Multi-Robot Systems Reinforcement Learning Intelligent Transportation Systems

Related papers