Application of intelligent autonomous agents to network fault management

Networks and distributed processing systems are of growing importance and have become an important substrate of modern information technology. As local area networks (LAN) increase in size and complexity, it becomes necessary to automate network management processes. Fault management is one of the functional areas of network management. It entails detection, identification and correction of anomalies that disrupt services on a network.;In this thesis, a framework is proposed for distributed computation and detection of LAN faults by introducing Intelligent Agents (IA). The IA agents make use of Bayesian networks as an intuitive modeling tool for bridging the gap between the wealth of information found in the Management Information Base (MIB) and the actual usage of such MIB variables in achieving fault identification. Decision making is handled by different distributed software entities, each with different capabilities and range of information. A Prototype reflecting the ideas discussed in this thesis is presented.