Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Agentic AI as a Network Control-Plane Intelligence Layer for Federated Learning over 6G

Loc X. Nguyen, Ji Su Yoon, Huy Q. Le, Yu Qiao, Avi Deb Raha, Eui-Nam Huh, Nguyen H. Tran, Zhu Han, Choong Seon Hong

TL;DR

An Agentic AI is proposed as the control layer for managing federated learning over 6G networks, which translates high-level task goals into actions that are aware of network conditions and uses closed-loop evaluation and memory to refine its decisions.

Abstract

The shift toward user-customized on-device learning places new demands on wireless systems: models must be trained on diverse, distributed data while meeting strict latency, bandwidth, and reliability constraints. To address this, we propose an Agentic AI as the control layer for managing federated learning (FL) over 6G networks, which translates high-level task goals into actions that are aware of network conditions. Rather than simply viewing FL as a learning challenge, our system sees it as a combined task of learning and network management. A set of specialized agents focused on retrieval, planning, coding, and evaluation utilizes monitoring tools and optimization methods to handle client selection, incentive structuring, scheduling, resource allocation, adaptive local training, and code generation. The use of closed-loop evaluation and memory allows the system to consistently refine its decisions, taking into account varying signal-to-noise ratios, bandwidth conditions, and device capabilities. Finally, our case study has demonstrated the effectiveness of the Agentic AI system's use of tools for achieving high performance.

Agentic AI as a Network Control-Plane Intelligence Layer for Federated Learning over 6G

TL;DR

An Agentic AI is proposed as the control layer for managing federated learning over 6G networks, which translates high-level task goals into actions that are aware of network conditions and uses closed-loop evaluation and memory to refine its decisions.

Abstract

The shift toward user-customized on-device learning places new demands on wireless systems: models must be trained on diverse, distributed data while meeting strict latency, bandwidth, and reliability constraints. To address this, we propose an Agentic AI as the control layer for managing federated learning (FL) over 6G networks, which translates high-level task goals into actions that are aware of network conditions. Rather than simply viewing FL as a learning challenge, our system sees it as a combined task of learning and network management. A set of specialized agents focused on retrieval, planning, coding, and evaluation utilizes monitoring tools and optimization methods to handle client selection, incentive structuring, scheduling, resource allocation, adaptive local training, and code generation. The use of closed-loop evaluation and memory allows the system to consistently refine its decisions, taking into account varying signal-to-noise ratios, bandwidth conditions, and device capabilities. Finally, our case study has demonstrated the effectiveness of the Agentic AI system's use of tools for achieving high performance.
Paper Structure (27 sections, 3 figures, 3 tables)

This paper contains 27 sections, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Agentic AI and Federated Learning
  3. Agentic AI Principles and Components
  4. Principles
  5. Components
  6. Federated Learning
  7. Basic Workflow
  8. Interpretable Processes
  9. Integration of Agentic AI for Federated Learning Workflow
  10. Planning Property: FL from the Viewpoint of Agentic AI
  11. Task Objective
  12. Client Selection and Incentive Design
  13. Local Training and Server Aggregation
  14. Wireless Communication and Resource Management
  15. Collaboration Property among Multi-Agent of Agentic AI
  16. ...and 12 more sections

Figures (3)

  • Figure 1: We provide an overview of the integration of the Agentic AI system into the FL framework, where we outline possible actions at each step to secure strong generalization performance and adapt to channel conditions.
  • Figure 2: The illustration of the autonomy of the Agentic AI system to train a deep learning model for a user-oriented task with the FL frameworks. Our proposed system is human-free, with specialized agents actively collaborating to train for the task.
  • Figure 3: Grouped bar chart comparing average SNR, average communication latency, and test accuracy across four client‑selection benchmarks (random, latency‑based, largest‑data, and class-diversity). The left y‑axis shows SNR/latency; the right y‑axis shows accuracy.