Soft Computing and Fault Management

Soft computing is a partnership between A.I. techniques that are tolerant of imprecision, uncertainty and partial truth, with the aim of obtaining a robust solution for complex systems. Telecommunication systems are built with extensive redundancy and complexity to ensure robustness and quality of service. To facilitate this requires complex fault identification and management systems. Fault identification and management is generally handled by reducing the amount of alarm events (symptoms) presented to the operating engineer through monitoring, filtering and masking. The ultimate goal is to determine and present the actual underlying fault. Fault Management is a complex task subject to uncertainty in the ’symptoms’ presented and as such is ideal for treatment by soft computing techniques. The aim of this paper is to present a soft computing approach to fault management in telecommunication systems. Two key approaches are considered; AI & soft computing rule discovery and techniques to attempt to present less symptoms with greater diagnostic assistance for the more traditional rule based system approach and a hybrid soft computing approach which utilises a genetic algorithm to learn Bayesian belief networks (BBNs). It is also highlighted that research and development of the two target Fault Management Systems are complementary.