Detection of Botnet traffic by using Neuro-fuzzy based Intrusion Detection

The attacks on various networks by intruders is on the rise and one of the most common model for launching attacks is by botnets. It has become very difficult for network administrators to detect and eradicate the bots in their network. Machine learning algorithms are increasingly being used for solving many classification problems including security systems for cloud. In this paper, we propose a new algorithm for the detection of botnet traffic by the use of neuro-fuzzy classification techniques. The dataset for the experimentation purpose was created by setting up an application on Eucalyptus cloud and attacking the application using various open source botnet simulation tools. The system achieved an accuracy of 94.78% with 15,000 instances and 56 attributes. The false positives of the system are considerably reduced when it is compared with the other related systems because of the introduction of the fuzzy rules into the system.