Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generative AI for Intent-Driven Network Management in 6G RAN: A Case Study on the Mamba Model

Md Arafat Habib, Medhat Elsayed, Yigit Ozcan, Pedro Enrique Iturria-Rivera, Majid Bavand, Melike Erol-Kantarci

TL;DR

The paper addresses the need for autonomous, intent-driven management in highly dynamic 6G RAN by introducing a novel Mamba-SSM-based architecture that covers intent processing, validation, and execution. It argues that replacing transformer-based components with selective memory Mamba models enables linear-time inference and better long-range reasoning within a hierarchical Open-RAN setup. The authors present a first-of-its-kind Mamba-driven end-to-end IDN framework, including Mamba- for intent interpretation, Mamba4Cast for predictive validation, and Decision Mamba for execution, demonstrating substantial improvements in QoS drift, throughput, energy efficiency, and decision latency in a multi-cell scenario. The work highlights practical deployment considerations and standardization implications, indicating strong potential for scalable GenAI-enabled management in disaggregated RAN and guiding future prototype validation on Open testbeds.

Abstract

With the emergence of 6G, mobile networks are becoming increasingly heterogeneous and dynamic, necessitating advanced automation for efficient management. Intent-Driven Networks (IDNs) address this by translating high-level intents into optimization policies. Large Language Models (LLMs) can enhance this process by understanding complex human instructions, enabling adaptive and intelligent automation. Given the rapid advancements in Generative AI (GenAI), a comprehensive survey of LLM-based IDN architectures in disaggregated Radio Access Network (RAN) environments is both timely and critical. This article provides such a survey, along with a case study on a selective State-Space Model (SSM)-enabled IDN architecture that integrates GenAI across three key stages: intent processing, intent validation, and intent execution. For the first time in the literature, we propose a hierarchical framework built on Mamba-SSM that introduces GenAI across all stages of the IDN pipeline. We further present a case study demonstrating that the proposed Mamba architecture significantly improves network performance through intelligent automation, surpassing existing IDN approaches. In a multi-cell 5G/6G scenario, the proposed architecture reduces quality of service drift by up to 70%, improves throughput by up to 80 Mbps, and lowers inference time to 60-70 ms, outperforming GenAI, reinforcement learning, and non-machine learning baselines.

Generative AI for Intent-Driven Network Management in 6G RAN: A Case Study on the Mamba Model

TL;DR

The paper addresses the need for autonomous, intent-driven management in highly dynamic 6G RAN by introducing a novel Mamba-SSM-based architecture that covers intent processing, validation, and execution. It argues that replacing transformer-based components with selective memory Mamba models enables linear-time inference and better long-range reasoning within a hierarchical Open-RAN setup. The authors present a first-of-its-kind Mamba-driven end-to-end IDN framework, including Mamba- for intent interpretation, Mamba4Cast for predictive validation, and Decision Mamba for execution, demonstrating substantial improvements in QoS drift, throughput, energy efficiency, and decision latency in a multi-cell scenario. The work highlights practical deployment considerations and standardization implications, indicating strong potential for scalable GenAI-enabled management in disaggregated RAN and guiding future prototype validation on Open testbeds.

Abstract

With the emergence of 6G, mobile networks are becoming increasingly heterogeneous and dynamic, necessitating advanced automation for efficient management. Intent-Driven Networks (IDNs) address this by translating high-level intents into optimization policies. Large Language Models (LLMs) can enhance this process by understanding complex human instructions, enabling adaptive and intelligent automation. Given the rapid advancements in Generative AI (GenAI), a comprehensive survey of LLM-based IDN architectures in disaggregated Radio Access Network (RAN) environments is both timely and critical. This article provides such a survey, along with a case study on a selective State-Space Model (SSM)-enabled IDN architecture that integrates GenAI across three key stages: intent processing, intent validation, and intent execution. For the first time in the literature, we propose a hierarchical framework built on Mamba-SSM that introduces GenAI across all stages of the IDN pipeline. We further present a case study demonstrating that the proposed Mamba architecture significantly improves network performance through intelligent automation, surpassing existing IDN approaches. In a multi-cell 5G/6G scenario, the proposed architecture reduces quality of service drift by up to 70%, improves throughput by up to 80 Mbps, and lowers inference time to 60-70 ms, outperforming GenAI, reinforcement learning, and non-machine learning baselines.

Paper Structure

This paper contains 14 sections, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. -Based Intent-driven Network Management Schemes
  3. State-of-the-art -Based Approaches for Management
  4. -Assisted Intent Processing and Management for Core Networks
  5. -Driven Multi-Agent and Negotiation-Based Intent Management
  6. End-to-End AI-Based Intent-Driven Network Automation
  7. Generative AI and Multimodal Intent-Based Network Management
  8. Open Research Directions for Enhancing Management
  9. Practical deployment considerations
  10. 3GPP and Open RAN standard implications
  11. Three-step Methodology for Management and its Integration with
  12. Hierarchical Integration with the three-step Management Architecture
  13. A Case Study on Mamba Model for Management
  14. Conclusions

Figures (5)

  • Figure 1: Integration of the proposed management framework with hierarchical architecture.
  • Figure 2: End-to-end GenAI framework for intent-driven network management.
  • Figure 3: Performance comparison of the proposed method against the baselines in terms of goal deviation (network delay and energy efficiency).
  • Figure 4: Performance comparison of the proposed method against the baselines in terms of throughput and reliability ( drift).
  • Figure 5: Performance comparison in terms of action inference time.