Context-aware Communication for Multi-agent Reinforcement Learning

TL;DR

This paper introduces CACOM, a context-aware communication protocol for multi-agent reinforcement learning that personalizes messages using attention mechanisms and reduces overhead with quantization, improving performance in bandwidth-limited scenarios.

Contribution

The paper proposes a novel two-stage, context-aware communication scheme with attention-based personalization and message quantization for MARL, addressing bandwidth limitations.

Findings

CACOM outperforms baseline methods in cooperative tasks under communication constraints.

Personalized messaging improves coordination and team performance.

Quantization effectively reduces communication overhead without sacrificing performance.

Abstract

Effective communication protocols in multi-agent reinforcement learning (MARL) are critical to fostering cooperation and enhancing team performance. To leverage communication, many previous works have proposed to compress local information into a single message and broadcast it to all reachable agents. This simplistic messaging mechanism, however, may fail to provide adequate, critical, and relevant information to individual agents, especially in severely bandwidth-limited scenarios. This motivates us to develop context-aware communication schemes for MARL, aiming to deliver personalized messages to different agents. Our communication protocol, named CACOM, consists of two stages. In the first stage, agents exchange coarse representations in a broadcast fashion, providing context for the second stage. Following this, agents utilize attention mechanisms in the second stage to selectively generate messages personalized for the receivers. Furthermore, we employ the learned step size quantization (LSQ) technique for message quantization to reduce the communication overhead. To evaluate the effectiveness of CACOM, we integrate it with both actor-critic and value-based MARL algorithms. Empirical results on cooperative benchmark tasks demonstrate that CACOM provides evident performance gains over baselines under communication-constrained scenarios. The code is publicly available at https://github.com/LXXXXR/CACOM.