Minimization of DDoS false alarm rate in Network Security; Refining fusion through correlation

Intrusion Detection Systems are designed to monitor a network environment and generate alerts whenever abnormal activities are detected. However, the number of these alerts can be very large making their evaluation a difficult task for a security analyst. Alert management techniques reduce alert volume significantly and potentially improve detection performance of an Intrusion Detection System. This thesis work presents a framework to improve the effectiveness and efficiency of an Intrusion Detection System by significantly reducing the false positive alerts and increasing the ability to spot an actual intrusion for Distributed Denial of Service attacks. Proposed sensor fusion technique addresses the issues relating the optimality of decision-making through correlation in multiple sensors framework. The fusion process is based on combining belief through Dempster Shafer rule of combination along with associating belief with each type of alert and combining them by using Subjective Logic based on Jøsang theory. Moreover, the reliability factor for any Intrusion Detection System is also addressed accordingly in order to minimize the chance of false diagnose of the final network state. A considerable number of simulations are conducted in order to determine the optimal performance of the proposed prototype