Using Neural Networks for Alarm Correlation in Cellular Phone Networks
Hermann Wietgrefe, K.-D. Tuchs, K. Jobmann, Guido Carls, Peter H. Fröhlich, Wolfgang Nejdl, Sebastian Steinfeld · 1999
Alarm handling and especially alarm correlation tools are necessary to manage large telecommunication networks. In this paper we describe our neural network based alarm correlation system, which uses a Cascade Correlation neural network to correlate alarms in a GSM network. The results of our approach called Cascade Correlation Alarm Correlator (CCAC) are shown. The behaviour in the case of noisy data is discussed and compared in detail to a codebook approach. Furthermore we contrast the neural network approach to another solution developed by our group which uses model-based diagnosis. 1 Introduction The main topic of this paper is fault management, especially alarm handling and alarm correlation in a GSM/DCS-mobile telephone network. Special attention is paid to the access network. The main problems addressed are alarm bursts, the task of alarm correlation and the development of tools to handle those bursts. 7.2.97, Seite 2 1.1 Structure of the GSM-Access-Network Mobile networks...