A Framework for Learning and Inference in Network Management

This paper presents a network management framework which builds the management information infrastructure and equips the management applications with learning and reasoning abilities for automatic and adaptive management tasks. Views are global virtual management information constructed via logical rules from the distributed physical management information. Through these views, management applications can access physical network entities. Management applications learn, network patterns and reason on the discovered patterns and pre-specified domain knowledge t o predict network behavior, diagnose problems, and trigger control actions. These abstract view definition, domain knowledge , and network patterns are a set of logical rules stored in the application-dependent M K B (Managem.ent Knowledge Base), while the physical management information is stored in the standard MIB (Management Information Base) at each node.