Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Large Language Model-Driven Cross-Domain Orchestration Using Multi-Agent Workflow

Xiaonan Xu, Haoshuo Chen, Jesse E. Simsarian, Roland Ryf, Nicolas K. Fontaine, Mikael Mazur, Lauren Dallachiesa, David T. Neilson

Abstract

We showcase an application that leverages multiple agents, powered by large language models and integrated tools, to collaboratively solve complex network operation tasks across various domains. The tasks include real-time topology retrieval, network optimization using physical models, and fiber switching facilitated by a robotic arm.

Large Language Model-Driven Cross-Domain Orchestration Using Multi-Agent Workflow

Abstract

We showcase an application that leverages multiple agents, powered by large language models and integrated tools, to collaboratively solve complex network operation tasks across various domains. The tasks include real-time topology retrieval, network optimization using physical models, and fiber switching facilitated by a robotic arm.

Paper Structure

This paper contains 2 sections, 2 figures.

Table of Contents

  1. Introduction
  2. Application Description

Figures (2)

  • Figure 1: Architecture and workflow of cross-domain network automation enabled by LLM-empowered agents. (DT: digital twin)
  • Figure 2: Output of multi-agent conversation involving aPlanner,Writer, andExecutor from the OTN domain (a) and the robotic domain (b), respectively. TheAdmininputs the task request. Responses highlighted in red signify the usage of tools such as function calling and code execution. Responses highlighted in purple exemplify the usage of the LLM's language skills such as data retrieval and simple mathematical reasoning. "[...]" indicates omitted information such as paths for saving files, code output, and descriptions of pre-defined functions. Theyellow boxsignifies omitted code or part of code generation.