Mining Sequential Alarm Patterns in a Telecommunication Database [chapter]

Pei-Hsin Wu, Wen-Chih Peng, Ming-Syan Chen
1999 Lecture Notes in Computer Science  
A telecommunication system produces daily a large amount of alarm data which contains hidden valuable information about the system behavior. The knowledge discovered from alarm data can be used in finding problems in networks and possibly in predicting severe faults. In this paper, we devise a solution procedure for mining sequential alarm patterns from the alarm data of a GSM system. First, by observing the features of the alarm data, we develop operations for data cleaning. Then, we transform
more » ... the alarm data into a set of alarm sequences. Note that the consecutive alarm events exist in the alarm sequences, and it is complicated to count the occurrence counts of events and extract patterns. Hence, we devise a new procedure to determine the occurrence count of the sequential alarm patterns in accordance with the nature of alarms. By utilizing time constraints to restrict the time difference between two alarm events, we devise a mining algorithm to discover useful sequential alarm patterns. The proposed mining algorithm is implemented and applied to test against a set of real alarm data provided by a cellular phone company. The quality of knowledge discovered is evaluated. The experimental results show that the proposed operations of data cleaning are able to improve the execution of our mining algorithm significantly and the knowledge obtained from the alarm data is very useful from the perspective of network operators for alarm prediction and alarm control. Due to recent technology advances, an increasing number of users are accessing various information system via wireless communication. Such information systems as stock trading, banking, wireless conferencing, are being provided by information services and application providers [5][6][7][13], and mobile users are able to access such information via wireless communication from anywhere at anytime [3][12][15].
doi:10.1007/3-540-45432-2_4 fatcat:erdo4ogsmvffvawsamllbj6xie