Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Beyond Self-Talk: A Communication-Centric Survey of LLM-Based Multi-Agent Systems

Bingyu Yan, Zhibo Zhou, Litian Zhang, Lian Zhang, Ziyi Zhou, Dezhuang Miao, Zhoujun Li, Chaozhuo Li, Xiaoming Zhang

TL;DR

<3-5 sentence high-level summary> This survey introduces a communication-centric framework for understanding LLM-based multi-agent systems, distinguishing system-level coordination from internal inter-agent communication. It surveys architectural families (Flat, Hierarchical, Team, Society, Hybrid), communication goals (Cooperation, Competition, Mixed), and emergent protocols (MCP, A2A, ANP), while also detailing internal strategies (One-by-One, Simultaneous-Talk, Summarizer) and paradigms (Message Passing, Speech Act, Blackboard). The authors identify critical challenges—scalability, security, interoperability, and benchmarking—and offer future directions toward unified protocols, multimodal coordination, and comprehensive evaluation suites. The work aims to guide researchers and practitioners in building robust, scalable, and secure LLM-MAS with coordinated, interpretable communication.

Abstract

Large language model-based multi-agent systems have recently gained significant attention due to their potential for complex, collaborative, and intelligent problem-solving capabilities. Existing surveys typically categorize LLM-based multi-agent systems (LLM-MAS) according to their application domains or architectures, overlooking the central role of communication in coordinating agent behaviors and interactions. To address this gap, this paper presents a comprehensive survey of LLM-MAS from a communication-centric perspective. Specifically, we propose a structured framework that integrates system-level communication (architecture, goals, and protocols) with system internal communication (strategies, paradigms, objects, and content), enabling a detailed exploration of how agents interact, negotiate, and achieve collective intelligence. Through an extensive analysis of recent literature, we identify key components in multiple dimensions and summarize their strengths and limitations. In addition, we highlight current challenges, including communication efficiency, security vulnerabilities, inadequate benchmarking, and scalability issues, and outline promising future research directions. This review aims to help researchers and practitioners gain a clear understanding of the communication mechanisms in LLM-MAS, thereby facilitating the design and deployment of robust, scalable, and secure multi-agent systems.

Beyond Self-Talk: A Communication-Centric Survey of LLM-Based Multi-Agent Systems

TL;DR

<3-5 sentence high-level summary> This survey introduces a communication-centric framework for understanding LLM-based multi-agent systems, distinguishing system-level coordination from internal inter-agent communication. It surveys architectural families (Flat, Hierarchical, Team, Society, Hybrid), communication goals (Cooperation, Competition, Mixed), and emergent protocols (MCP, A2A, ANP), while also detailing internal strategies (One-by-One, Simultaneous-Talk, Summarizer) and paradigms (Message Passing, Speech Act, Blackboard). The authors identify critical challenges—scalability, security, interoperability, and benchmarking—and offer future directions toward unified protocols, multimodal coordination, and comprehensive evaluation suites. The work aims to guide researchers and practitioners in building robust, scalable, and secure LLM-MAS with coordinated, interpretable communication.

Abstract

Large language model-based multi-agent systems have recently gained significant attention due to their potential for complex, collaborative, and intelligent problem-solving capabilities. Existing surveys typically categorize LLM-based multi-agent systems (LLM-MAS) according to their application domains or architectures, overlooking the central role of communication in coordinating agent behaviors and interactions. To address this gap, this paper presents a comprehensive survey of LLM-MAS from a communication-centric perspective. Specifically, we propose a structured framework that integrates system-level communication (architecture, goals, and protocols) with system internal communication (strategies, paradigms, objects, and content), enabling a detailed exploration of how agents interact, negotiate, and achieve collective intelligence. Through an extensive analysis of recent literature, we identify key components in multiple dimensions and summarize their strengths and limitations. In addition, we highlight current challenges, including communication efficiency, security vulnerabilities, inadequate benchmarking, and scalability issues, and outline promising future research directions. This review aims to help researchers and practitioners gain a clear understanding of the communication mechanisms in LLM-MAS, thereby facilitating the design and deployment of robust, scalable, and secure multi-agent systems.

Paper Structure

This paper contains 49 sections, 9 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Background
  3. LLM-Based Agents
  4. Traditional Multi-Agent Systems
  5. LLM-Based Multi-Agent Systems
  6. System-Level Communication
  7. Communication Architecture
  8. Flat Architecture
  9. Hierarchical Architecture
  10. Team Architecture.
  11. Society Architecture
  12. Hybrid Architecture
  13. Communication Goal
  14. Cooperation
  15. Competition
  16. ...and 34 more sections

Figures (9)

  • Figure 1: Position of this survey within existing LLM-MAS literature. The figure contrasts prior general and domain-specific surveys with the communication-centric perspective, underscoring the need for a cross-cutting framework that spans tasks, architectures, workflows, and interaction between agents
  • Figure 2: The structure of this paper
  • Figure 3: Five canonical communication architectures for LLM-MAS
  • Figure 4: Communication goals driving multi-agent interaction
  • Figure 5: Emerging communication protocols for LLM-MAS
  • ...and 4 more figures