CAIC: Congestion-Aware Intent Communication for Multi-Agent Reinforcement Learning

Pei Li, Yongkang Zhang, Zhonglin Lv, Jinmin Zhu, Jiangjin Yin
Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, PMLR 337:3705-3718, 2026.

Abstract

In multi-agent reinforcement learning, communication is essential for effective cooperation. Most existing methods assume instantaneous message delivery, and the few delay-aware studies consider only external fixed or stochastic delays, neglecting queueing delays caused by competition for a shared channel. We formulate the shared channel as a queueing system with state-dependent service rates, and propose Congestion-Aware Intent Communication ({CAIC}). To ensure message validity under delay, {CAIC} employs a temporal masked autoencoder to predict each agent’s future trajectory and encode it into a delay-robust intent message, with attention-based fusion integrating asynchronous messages at the receiver. To maintain timeliness, {CAIC} dynamically adjusts communication frequency based on intent message changes and delay estimates, proactively mitigating congestion. Experiments on Hallway, MPE, and SMAC demonstrate that {CAIC} outperforms existing baselines on most evaluated scenarios in both no-delay and queueing-delay settings, and ablation studies reveal that stale messages under delay can be even more harmful than no communication.

Cite this Paper

BibTeX
@InProceedings{pmlr-v337-li26h, title = {{CAIC}: Congestion-Aware Intent Communication for Multi-Agent Reinforcement Learning}, author = {Li, Pei and Zhang, Yongkang and Lv, Zhonglin and Zhu, Jinmin and Yin, Jiangjin}, booktitle = {Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence}, pages = {3705--3718}, year = {2026}, editor = {Perković, Emilija and Malinsky, Daniel}, volume = {337}, series = {Proceedings of Machine Learning Research}, month = {17--21 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v337/main/assets/li26h/li26h.pdf}, url = {https://proceedings.mlr.press/v337/li26h.html}, abstract = {In multi-agent reinforcement learning, communication is essential for effective cooperation. Most existing methods assume instantaneous message delivery, and the few delay-aware studies consider only external fixed or stochastic delays, neglecting queueing delays caused by competition for a shared channel. We formulate the shared channel as a queueing system with state-dependent service rates, and propose Congestion-Aware Intent Communication ({CAIC}). To ensure message validity under delay, {CAIC} employs a temporal masked autoencoder to predict each agent’s future trajectory and encode it into a delay-robust intent message, with attention-based fusion integrating asynchronous messages at the receiver. To maintain timeliness, {CAIC} dynamically adjusts communication frequency based on intent message changes and delay estimates, proactively mitigating congestion. Experiments on Hallway, MPE, and SMAC demonstrate that {CAIC} outperforms existing baselines on most evaluated scenarios in both no-delay and queueing-delay settings, and ablation studies reveal that stale messages under delay can be even more harmful than no communication.} }
Endnote
%0 Conference Paper %T CAIC: Congestion-Aware Intent Communication for Multi-Agent Reinforcement Learning %A Pei Li %A Yongkang Zhang %A Zhonglin Lv %A Jinmin Zhu %A Jiangjin Yin %B Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2026 %E Emilija Perković %E Daniel Malinsky %F pmlr-v337-li26h %I PMLR %P 3705--3718 %U https://proceedings.mlr.press/v337/li26h.html %V 337 %X In multi-agent reinforcement learning, communication is essential for effective cooperation. Most existing methods assume instantaneous message delivery, and the few delay-aware studies consider only external fixed or stochastic delays, neglecting queueing delays caused by competition for a shared channel. We formulate the shared channel as a queueing system with state-dependent service rates, and propose Congestion-Aware Intent Communication ({CAIC}). To ensure message validity under delay, {CAIC} employs a temporal masked autoencoder to predict each agent’s future trajectory and encode it into a delay-robust intent message, with attention-based fusion integrating asynchronous messages at the receiver. To maintain timeliness, {CAIC} dynamically adjusts communication frequency based on intent message changes and delay estimates, proactively mitigating congestion. Experiments on Hallway, MPE, and SMAC demonstrate that {CAIC} outperforms existing baselines on most evaluated scenarios in both no-delay and queueing-delay settings, and ablation studies reveal that stale messages under delay can be even more harmful than no communication.
APA
Li, P., Zhang, Y., Lv, Z., Zhu, J. & Yin, J.. (2026). CAIC: Congestion-Aware Intent Communication for Multi-Agent Reinforcement Learning. Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 337:3705-3718 Available from https://proceedings.mlr.press/v337/li26h.html.

Related Material