AgentRAN: An Agentic AI Architecture for Autonomous Control of Open 6G Networks

TL;DR

AgentRAN introduces an AI-native, Open RAN-aligned agentic framework that translates natural language intents into autonomous network actions via a hierarchical fabric of LLM-powered agents. It decomposes intents across three dimensions—timescales, geography, and protocol layers—and coordinates agents using context aggregation, intent cascading, and constraint propagation, all under a central AI-RAN Manager. A key innovation is the AI-RAN Factory, which uses data lakes to perform continuous learning, generating improved agents through code generation, distillation, fine-tuning, and hybrid creation, validated by sandbox testing before deployment. Real-world 5G experiments demonstrate rapid, auditable adaptation to operator intents (e.g., uplink scheduling and power) and autonomous self-improvement under distribution shift, signaling a pathway toward networks that evolve their own intelligence while maintaining transparency and safety margins.

Abstract

Despite the programmable architecture of Open RAN, today's deployments still rely heavily on static control and manual operations. To move beyond this limitation, we introduce AgentRAN, an AI-native, Open RAN-aligned agentic framework that generates and orchestrates a fabric of distributed AI agents based on natural language intents. Unlike traditional approaches that require explicit programming, AgentRAN's LLM-powered agents interpret natural language intents, negotiate strategies through structured conversations, and orchestrate control loops across the network. AgentRAN instantiates a self-organizing hierarchy of agents that decompose complex intents across time scales (from sub-millisecond to minutes), spatial domains (cell to network-wide), and protocol layers (PHY/MAC to RRC). A central innovation is the AI-RAN Factory, which continuously generates improved agents and algorithms from operational data, transforming the network into a system that evolves its own intelligence. We validate AgentRAN through live 5G experiments, demonstrating dynamic adaptation to changing operator intents across power control and scheduling. Key benefits include transparent decision-making (all agent reasoning is auditable), bootstrapped intelligence (no initial training data required), and continuous self-improvement via the AI-RAN Factory.

Figures (3)

• Figure 1: AgentRAN high-level architecture.
• Figure 2: Impact of agent decisions on UEs during normal phase ($< 300$ s), emergency ($300$-$600$ s), and post-emergency ($> 600$ s).
• Figure 3: Propsim-based interference prediction scenario.