Intent-based management for next-generation networks: an LLM-centric approach

Intent-based networking (IBN) is crucial in enabling autonomous networks by specifying goals and constraints at a higher level to the Network Management System. TMForum has specified a dedicated architecture and model that rely on Intents to handle and manage communication services, paving the way for autonomous systems towards 6G. Intents that represent an abstracted operational goal specified by the communication service owner, which is usually provided as input to the Network Management System (NMS). The latter, in turn, generates the necessary low-level configurations to fulfill these Intents. The current model of expressing Intents still requires significant effort in writing JSON and YAML structures, demanding a detailed comprehension of the format and model specified by the Northbound Interface (NBI). This process is sometimes not straightforward, and adhering to the structure of these NBIs takes time. A natural evolution for IBN is to move beyond human-readable languages and transition towards natural language. In this talk, we will discuss the evolution of Intents in 6G relaying Large Language Model (LLM) that translates human language into operational intents to deploy communication systems, leveraging few-shot learning and human-in-the-loop Feedback.