Network expert diagnostic system for real-time control
Terry L. Janssen · 1989
Data communications networks are controlled by network management systems that are responsible for performance and fault management. This paper presents an expert system capable of performing fault and performance management through different levels of autonomous control. A blackboard architecture design provides for processing of multiple lines of machine reasoning and planning: the set of all unresolved events is used to generate hypotheses of network state through event correlation and ancillary network information; supporting network data provides evidence that supports or refutes the hypotheses; conclusions are drawn from the hypotheses; plans of corrective action are built, executed, and monitored to attempt improvement to a “normal” network state.