Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-Agentic AI for Conflict-Aware rApp Policy Orchestration in Open RAN

Haiyuan Li, Yulei Wu, Dimitra Simeonidou

TL;DR

This work proposes a Multi-Agentic AI framework to automate rApp policy generation and orchestration in Open RAN, and achieves over 70% improvement in deployment accuracy and 95% reduction in reasoning cost compared to baseline methods.

Abstract

Open Radio Access Network (RAN) enables flexible, AI-driven control of mobile networks through disaggregated, multi-vendor components. In this architecture, xApps handle real-time functions, whereas rApps in the non-real-time controller generate strategic policies. However, current rApp development remains largely manual, brittle, and poorly scalable as xApp diversity proliferates. In this work, we propose a Multi-Agentic AI framework to automate rApp policy generation and orchestration. The architecture integrates three specialized large language model (LLM)-based agents, Perception, Reasoning, and Refinement, supported by retrieval-augmented generation (RAG) and memory-based analogical reasoning. These agents collectively analyze potential conflicts, synthesize intent-aligned control pipelines, and incrementally refine deployment decisions. Experiments across diverse deployment scenarios demonstrate that the proposed system achieves over 70% improvement in deployment accuracy and 95% reduction in reasoning cost compared to baseline methods, while maintaining zero-shot generalization to unseen intents. These results establish a scalable and conflict-aware solution for fully autonomous, zero-touch rApp orchestration in Open RAN.

Multi-Agentic AI for Conflict-Aware rApp Policy Orchestration in Open RAN

TL;DR

This work proposes a Multi-Agentic AI framework to automate rApp policy generation and orchestration in Open RAN, and achieves over 70% improvement in deployment accuracy and 95% reduction in reasoning cost compared to baseline methods.

Abstract

Open Radio Access Network (RAN) enables flexible, AI-driven control of mobile networks through disaggregated, multi-vendor components. In this architecture, xApps handle real-time functions, whereas rApps in the non-real-time controller generate strategic policies. However, current rApp development remains largely manual, brittle, and poorly scalable as xApp diversity proliferates. In this work, we propose a Multi-Agentic AI framework to automate rApp policy generation and orchestration. The architecture integrates three specialized large language model (LLM)-based agents, Perception, Reasoning, and Refinement, supported by retrieval-augmented generation (RAG) and memory-based analogical reasoning. These agents collectively analyze potential conflicts, synthesize intent-aligned control pipelines, and incrementally refine deployment decisions. Experiments across diverse deployment scenarios demonstrate that the proposed system achieves over 70% improvement in deployment accuracy and 95% reduction in reasoning cost compared to baseline methods, while maintaining zero-shot generalization to unseen intents. These results establish a scalable and conflict-aware solution for fully autonomous, zero-touch rApp orchestration in Open RAN.
Paper Structure (13 sections, 2 equations, 4 figures)

This paper contains 13 sections, 2 equations, 4 figures.

Table of Contents

  1. Introduction
  2. xApp coordination in Open RAN: System context and Problem formulation
  3. Multi-Agentic AI for rApp policy generation
  4. Perception Agent
  5. Reasoning Agent
  6. Knowledge Retrieval and Memory-Augmented Reasoning
  7. Refinement Agent
  8. Experiment and results
  9. Experimental setup
  10. Open RAN environment
  11. Agentic AI
  12. Results
  13. Conclusion

Figures (4)

  • Figure 1: An Open RAN scenario where high-level RAN intents are translated into rApp policies, each instantiated as a pipeline of xApps managing real-time network attributes.
  • Figure 2: The Multi-Agentic AI framework.
  • Figure 3: Evaluation of policy generation accuracy and conflict-aware deployment success rate (While several baselines achieve comparable accuracy, they may differ significantly in reasoning efficiency (see Fig. \ref{['fig:attempts']}).).
  • Figure 4: Evaluation of orchestration efficiency: attempts required for optimal policy generation and deployment.