Intelligent network management: a heterogeneous knowledge source approach

The knowledge and data engineering aspects of modeling the structural and functional components of a modern telecommunications network are presented. The networks are highly intelligent and dynamic, with the capability to reroute traffic between switches automatically and to support a customer's virtual network. The goal of this modeling is to automatically identify and isolate network faults as manifested by alarms given off by network elements. The authors posit the need for several knowledge/data sources that cooperate in this process: (1) the network topology and connectivity model, (2) the alarm monitoring model, and (3) the rule and case base to provide high-level problem-solving guidance to isolate primary alarms (faults) from their triggered secondary alarms. Models (1) and (2) are developed in detail, and it is shown that they can be merged to facilitate the correlation of alarm events to network elements. Both models are specified using the knowledge entity relationship model. >