RAG-Enabled Intent Reasoning for Application-Network Interaction

TL;DR

This paper tackles the challenge of translating diverse, domain-specific application intents into low-level network configurations within intent-based networking (IBN). It introduces the Intent-RAG framework, which combines domain-context knowledge bases, retrieval augmented generation, machine reasoning, and few-shot prompting to translate high-level, potentially non-technical intents into structured network intents compliant with SDOs. Through implementation and evaluation against vanilla-RAG and no-RAG baselines, Intent-RAG demonstrates improved retrieval quality and higher fidelity in generated network intents, while maintaining acceptable translation times. The approach offers a practical path to scalable, privacy-preserving cross-domain intent translation, enabling non-expert users to influence network behavior with up-to-date knowledge bases and structured policy pipelines.

Abstract

Intent-based network (IBN) is a promising solution to automate network operation and management. IBN aims to offer human-tailored network interaction, allowing the network to communicate in a way that aligns with the network users' language, rather than requiring the network users to understand the technical language of the network/devices. Nowadays, different applications interact with the network, each with its own specialized needs and domain language. Creating semantic languages (i.e., ontology-based languages) and associating them with each application to facilitate intent translation lacks technical expertise and is neither practical nor scalable. To tackle the aforementioned problem, we propose a context-aware AI framework that utilizes machine reasoning (MR), retrieval augmented generation (RAG), and generative AI technologies to interpret intents from different applications and generate structured network intents. The proposed framework allows for generalized/domain-specific intent expression and overcomes the drawbacks of large language models (LLMs) and vanilla-RAG framework. The experimental results show that our proposed intent-RAG framework outperforms the LLM and vanilla-RAG framework in intent translation.

Figures (3)

• Figure 1: Intent-RAG Framework for Intent Interpreting.
• Figure 2: Retrieval part performance of vanilla-RAG and intent-RAG.
• Figure 3: Generative part performance of vanilla-RAG and intent-RAG.