Towards Adaptive, Scalable, and Robust Coordination of LLM Agents: A Dynamic Ad-Hoc Networking Perspective

TL;DR

This paper introduces RAPS, a reputation-aware publish-subscribe system that enables adaptive, scalable, and robust coordination among large language model agents by leveraging dynamic networking principles and reputation management.

Contribution

The paper proposes RAPS, a novel coordination framework for LLM agents that combines publish-subscribe communication, dynamic intent refinement, and Bayesian reputation for robustness.

Findings

RAPS achieves effective scalability in multi-agent systems.

The reputation mechanism detects and isolates malicious agents.

Experiments demonstrate improved adaptivity and robustness across benchmarks.

Abstract

Multi-agent architectures built on large language models (LLMs) have demonstrated the potential to realize swarm intelligence through well-crafted collaboration. However, the substantial burden of manual orchestration inherently raises an imperative to automate the design of agentic workflows. We frame such an agent coordination challenge as a classic problem in dynamic ad-hoc networking: How to establish adaptive and reliable communication among a scalable number of agentic hosts? In response to this unresolved dilemma, we introduce RAPS, a reputation-aware publish-subscribe paradigm for adaptive, scalable, and robust coordination of LLM agents. RAPS is grounded in the Distributed Publish-Subscribe Protocol, allowing LLM agents to exchange messages based on their declared intents rather than predefined topologies. Beyond this substrate, RAPS further incorporates two coherent overlays: (i) Reactive Subscription, enabling agents to dynamically refine their intents; and (ii) Bayesian Reputation, empowering each agent with a local watchdog to detect and isolate malicious peers. Extensive experiments over five benchmarks showcase that our design effectively reconciles adaptivity, scalability, and robustness in a unified multi-agent coordination framework.