Proceedings of the International Workshop on Applications of Neural Networks to Telecommunications

Using Distributed Neural Networks to Identify Faults in Switching Systems

Tadashi Sone

NTT Communication Switching Labs,9-11, Midori-Cho 3-Chome, Musashino-Shi,Tokyo 180 [e-mail]

Abstract

A fault identification system that employs a distributed neural network architecture for switching system diagnosis is described. This distributed neural network structure consists of several micro-viewing networks and one macro-viewing network, that take into account the diagnosis control flow and switching system architecture. The module for integrating the output of these distributed networks employs an expert system. This diagnosis system provides a higher diagnosis rate than a single neural network.

1  Introduction

Electronic switching systems(ESSs) ...