FAULT MANAGEMENT IN COMMUNICATION NETWORKS USING PROBABILISTIC REASONING AND BELIEF MODELS

The challenge to providing an efficient and reliable fault management system in cellular networks demands intelligent fault detection and correction procedures. Recent investigative study in CDMA neiworks in Nigeria shows that the frequent faults include: faulty links, radio frequency (RI') loss and trunk outages, which gradually degrade quality of service of the networks. Most of these faults occur in the El link between Wit and the BSC. Analysis of logged data over a period of one year shows that signal out of step occurs with probability of 0.62, unusable link occurs with 0.13, while other faults occurs with 0.25. This paper presents a system designed to improve the present system of fault detection and isolation in the base station subsystem of a CDMA network based on this abnormalities. It is based on an intelligent agent technology, with probabilistic networks as framework for knowledge representation and evidence inference. It was found to be robust and has 70% reliability.