Lightweight Natural Language Driven Intent Translation Mechanism for Intent Based Networking

Intent-based networking (IBN) simplifies tedious network configuration. It allows users without network expertise to configure the network. Users only need to care about what is needed, without describing its implementation. This paper proposes a lightweight natural language-driven intent translation mechanism. This mechanism realizes the translation and delivery of user intent at multiple service levels. Compared with the existing intent translation mechanism, the advantages of this mechanism include the following three points: (1) It depends on flexible natural language and is not limited to a specific structure. (2) It does not require excessive user configuration, which uses the network topology collected by the ONOS controller to automatically configure network parameters. (3) It has a learning function. As the translation work progresses, the knowledge base is continuously supplemented to improve the translation accuracy. In our experimental environment and dataset, the intent translation mechanism has a high translation accuracy rate, and the average translation time remains around 0.02s when the input set size is 200 bytes.