I2BN: Intelligent Intent Based Networks

Abstract

Intent based network management reduces the complexity of network programming from a growing set of deeply technical APIs to context-free high-level objectives that the network should autonomously achieve and keep. The practical implementation of an intent based network requires substantial automation technology embedded in the network. Automation should cover the entire lifecycle of intents, from their ingestion to fulfillment and assurance. This article investigates the feasibility of automatically assembling interworking implementation units into intent specific automation pipelines, where units are reusable self-learning closed loop micro-services with self-declared capabilities. Each closed loop may gain knowledge and respond to dynamically changing network conditions, thereby enabling network autonomy in reaching the declared intent objectives. The human-network intent interface for expressing intents is proposed to be based on the aggregation of the deployed network and service automation capabilities, rather than a formalism decoupled from the actual network implementation. This principle removes the ambiguity and compatibility gap between human intent definition and machine intent fulfillment, while retaining the flexibility and extendibility of the intents offered by any specific system via onboarding additional micro-services with novel capabilities. The concepts discussed by the article fit into the architecture and closed loop work items already defined by ETSI ZSM and provides considerations towards new areas such as intent driven autonomous networks and enablers for automation.