Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Intent Assurance using LLMs guided by Intent Drift

Kristina Dzeparoska, Ali Tizghadam, Alberto Leon-Garcia

TL;DR

The paper tackles the problem of maintaining alignment between intended network behaviors and real-time operation in Intent-Based Networking by introducing an assurance framework that detects and mitigates intent drift. It formalizes intents as KPI vectors, uses KPI quantization and logical aggregation to assess health, and leverages LLMs to both fulfill intents and generate corrective assurance actions when drift is detected. A key contribution is the formalization of drift measurement via $\Delta \overrightarrow{K}$, the distance $\text{Distance}$, and the gradient $\nabla E$, which guide resource and service-level adjustments; this is integrated into a progressive LLM pipeline with feedback. The framework is demonstrated through a use-case on NetFlow collectors, showing that LLMs can decompose intents into policies, perform fulfillment, and execute assurance actions with measurable time savings. The work points to practical impact in automating IBN operations, supporting rapid, AI-guided self-healing and adaptation in dynamic networks, and highlights directions for future enhancements such as vector-db in-context learning and digital twins for policy validation.

Abstract

Intent-Based Networking (IBN) presents a paradigm shift for network management, by promising to align intents and business objectives with network operations--in an automated manner. However, its practical realization is challenging: 1) processing intents, i.e., translate, decompose and identify the logic to fulfill the intent, and 2) intent conformance, that is, considering dynamic networks, the logic should be adequately adapted to assure intents. To address the latter, intent assurance is tasked with continuous verification and validation, including taking the necessary actions to align the operational and target states. In this paper, we define an assurance framework that allows us to detect and act when intent drift occurs. To do so, we leverage AI-driven policies, generated by Large Language Models (LLMs) which can quickly learn the necessary in-context requirements, and assist with the fulfillment and assurance of intents.

Intent Assurance using LLMs guided by Intent Drift

TL;DR

The paper tackles the problem of maintaining alignment between intended network behaviors and real-time operation in Intent-Based Networking by introducing an assurance framework that detects and mitigates intent drift. It formalizes intents as KPI vectors, uses KPI quantization and logical aggregation to assess health, and leverages LLMs to both fulfill intents and generate corrective assurance actions when drift is detected. A key contribution is the formalization of drift measurement via , the distance , and the gradient , which guide resource and service-level adjustments; this is integrated into a progressive LLM pipeline with feedback. The framework is demonstrated through a use-case on NetFlow collectors, showing that LLMs can decompose intents into policies, perform fulfillment, and execute assurance actions with measurable time savings. The work points to practical impact in automating IBN operations, supporting rapid, AI-guided self-healing and adaptation in dynamic networks, and highlights directions for future enhancements such as vector-db in-context learning and digital twins for policy validation.

Abstract

Intent-Based Networking (IBN) presents a paradigm shift for network management, by promising to align intents and business objectives with network operations--in an automated manner. However, its practical realization is challenging: 1) processing intents, i.e., translate, decompose and identify the logic to fulfill the intent, and 2) intent conformance, that is, considering dynamic networks, the logic should be adequately adapted to assure intents. To address the latter, intent assurance is tasked with continuous verification and validation, including taking the necessary actions to align the operational and target states. In this paper, we define an assurance framework that allows us to detect and act when intent drift occurs. To do so, we leverage AI-driven policies, generated by Large Language Models (LLMs) which can quickly learn the necessary in-context requirements, and assist with the fulfillment and assurance of intents.
Paper Structure (21 sections, 18 equations, 1 figure, 4 tables)

This paper contains 21 sections, 18 equations, 1 figure, 4 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Intent Assurance Modelling
  4. Formalizing intent with KPIs
  5. Few-shot learning to formalize intents and extract KPIs
  6. KPI Quantization
  7. Logical KPI Assessment
  8. Detecting KPI Deviations and measuring Intent Drift
  9. KPI Assurance Assessment
  10. Resource Health ($h_r$)
  11. Composite Resource Health ($k_{H_{r}}$)
  12. Service Health ($k_{H_s}$)
  13. Service Availability ($k_{A_s}$)
  14. Intent Health
  15. Use-case Evaluation
  16. ...and 6 more sections

Figures (1)

  • Figure 1: Snippet of the assurance analysis by the LLM.