Model-Based-Diagnosis for Fault Management in Telecommunications Networks

Abstract

Fault management is a crucial problem for telecommunica-tion networks. The network complexity requires articial intelligence techniques to assist the operators in supervision tasks. Initially expert systems techniques were proposed, presently various techniques are used: neural networks, constraint satisfaction problems, Petri networks. These methods are based on the analysis of the breakdown situa-tions observed in the real system, consequently, they have the same problem as expert systems: they are enable to reason about new faults and the level of explanation of fault is poor. The continual change of the telecommunication networks reduces the effectiveness of these methods. The model based techniques propose a framework based on the adaptive modeling of the system and the prediction of the behavior of