Ontology-Based Network Intent Refinement Framework

Abstract

Intent-driven network (IDN, also named as intent-based network, IBN) allows users to implement policies in a declarative way, instead of treating complete them in the underlying network. However, how to refine intent in natural language remains to be further researched. Although there exist emerging works on intent refinement, the problem of heterogeneous forms of intents in multiple carrier networks is still unsolved at present. As a formalized term set, ontology makes it possible to unify the various terms at the semantic level. The ontology-based architecture of IDN consists of the application layer, the intent-enabled layer, and the infrastructure layer. Users can input intent in the application layer in various ways, and the input intent can be converted into an intent tuple as . The intent-enabled layer constructs an intent ontology, on the basis of intent tuple and network configuration protocol (NETCONF) information from the infrastructure layer. In addition, the refined intent can be verified both in natural language processing (NLP) level and the semantic level. In this paper, we design an intelligent intent refinement system to accomplish intent refinement.