AGORA: Agentic Green Orchestration Architecture for Beyond 5G Networks

TL;DR

AGORA introduces an agentic, local LLM-driven closed-loop architecture that grounds sustainability intents in real-time telemetry and directly actuates UPF routing to optimize energy efficiency in Beyond-5G networks. By evaluating multiple open LLM backends within a MEC testbed, the study demonstrates a pronounced latency-energy coupling in tool-driven control loops and shows that compact models can achieve low energy footprints while enabling correct policy execution, including non-zero migration under MEC stress. The work provides a practical methodology for policy compliance, telemetry integration, and end-to-end evaluation of sustainability-oriented orchestration, highlighting the importance of model choice for energy efficiency and policy fidelity. The findings support sustainability-first, intent-driven network operations and point to future enhancements such as standardized interfaces, broader model exploration, and incorporation of carbon intensity signals for safer, scalable deployment.

Abstract

Effective management and operational decision-making for complex mobile network systems present significant challenges, particularly when addressing conflicting requirements such as efficiency, user satisfaction, and energy-efficient traffic steering. The literature presents various approaches aimed at enhancing network management, including the Zero-Touch Network (ZTN) and Self-Organizing Network (SON); however, these approaches often lack a practical and scalable mechanism to consider human sustainability goals as input, translate them into energy-aware operational policies, and enforce them at runtime. In this study, we address this gap by proposing the AGORA: Agentic Green Orchestration Architecture for Beyond 5G Networks. AGORA embeds a local tool-augmented Large Language Model (LLM) agent in the mobile network control loop to translate natural-language sustainability goals into telemetry-grounded actions, actuating the User Plane Function (UPF) to perform energy-aware traffic steering. The findings indicate a strong latency-energy coupling in tool-driven control loops and demonstrate that compact models can achieve a low energy footprint while still facilitating correct policy execution, including non-zero migration behavior under stressed Multi-access Edge Computing (MEC) conditions. Our approach paves the way for sustainability-first, intent-driven network operations that align human objectives with executable orchestration in Beyond-5G infrastructures.

Figures (16)

• Figure 1: Agentic closed-loop sustainability orchestration in a 5G MEC testbed.
• Figure 2: Total active MEC energy across all runs.
• Figure 3: CDF of agent inference latency.
• Figure 4: Average energy per generated token.
• Figure 5: Total generated tokens across runs.