Defining the threshold bounds for effective modelling of aggregated IDS.

The idea behind the aggregation of multiple intrusion detections (IDS) is due to the inability of stand-alone IDS to meet the demand of today's network traffic with high complexity of intrusions. Multiple IDS can be aggregated in order to meet the requirement for low false alarms and the demand for high detection rate. This paper shows that by appropriate adjustment in the threshold results in better performance during the aggregation process of the multiple IDS. Complementary IDS are used to increase the detection rate rather than the individual IDS. A number of evaluation metrics is used to show that there is a significant enhancement in the performance of the IDS system. The threshold bounds are derived using three well known inequalities in probability theory and compared for better performance. [ABSTRACT FROM AUTHOR]