LLM-Based Multi-Agent Systems over Wireless Networks: A Joint Agent--Network Design Perspective

2026-09-29

AI Summary

This study addresses the performance bottlenecks and trust discontinuities in LLM-based multi-agent systems over wireless edge networks, which arise from the coupling between inference dependencies and network resources. To overcome the limitations of conventional decoupled architectures, this work proposes a joint agent-network design framework that pioneers the integration of information flow control with network-verified traceability. By co-optimizing interaction scheduling, message transmission, and topology configuration, it enables efficient collaborative reasoning under communication constraints. Methodologically, the approach combines joint resource allocation, synergistic message selection-transmission design, and blockchain-based provenance within a V2X simulation environment. Experimental results demonstrate that the proposed scheme significantly outperforms independently designed baselines in task completion rates under resource-constrained conditions.

Abstract

As large language models (LLMs) evolve from standalone models into collaborative agents embedded in physical systems, their reasoning and execution are increasingly distributed across wireless edge nodes. In this setting, wireless networks are experiencing a paradigm shift from only providing data connectivity to supporting the multi-agent reasoning workflow itself. The task performance of such network-constrained LLM-based multi-agent systems (MASs) is jointly affected by the multi-agent reasoning dependencies as well as the underlying network connectivity and edge resources. This coupling gives rise to various technical challenges, including the metric misalignment and message redundancy, state inconsistency and topology mismatch, as well as resource limitation and trust discontinuity. To address these challenges, this article develops a novel joint agent--network design perspective that coordinates decisions on both sides of the system. Specifically, we present the joint design of agent--interaction scheduling and resource allocation, the message selection-transmission co-design, as well as the joint agent--network topology design and workload--resource allocation. Furthermore, we consider the network-verified provenance that is linked with agent-side information-flow control to constrain how received information affects subsequent operations. An illustrative vehicle-to-everything (V2X) case study shows that jointly adapting agent-side interaction decisions and network operations improves task completion under communication and edge-resource constraints, outperforming the conventional agent-only and wireless-only separate designs.

Problem

• Large Language Models
• Multi-Agent Systems
• Wireless Networks
• Edge Computing
• Joint Agent-Network Design

Innovation

• Joint Agent-Network Design
• LLM-Based Multi-Agent Systems
• Resource Allocation
• Message Co-design
• Network-Verified Provenance