DMO-GPT: An Intent-Driven Framework for Distributed 6G Management and Orchestration

With the increasing demands for high quality of service (QoS) in 6G networks, managing these complex systems requires intelligent Operations Support Systems (OSSs). Standardization institutions such as ETSI and 3GPP mandate that OSSs enable end-to-end, cross-domain management across all 6G components. To this end, ongoing initiatives within these institutions are actively defining API frameworks to simplify interactions with OSSs, focusing on Intent-Based Networking (IBN) and Zero-touch network and Service Management (ZSM) APIs. However, the diversity of API standards across different OSSs presents significant challenges when managing 6G use cases across multiple Mobile Network Operators (MNOs). To this end, we propose in this article a novel framework that leverages Large Language Models (LLMs) to enable natural language-based interactions with OSSs across multiple MNOs. To address the complexity of fulfilling user intents, which may involve multiple low-level API calls across heterogeneous OSSs, the proposed solution integrates multi-agent LLMs with an hierarchical planning mechanism. The framework offers two key advantages: a simple natural language-based interaction; and an adaptive system capable of autonomously accommodating new OSS (API) features. Real-world experiments validate the efficacy of the proposed framework, demonstrating its ability to efficiently manage diverse 6G OSSs and enhance the accessibility, interoperability, and automation of 6G network management.