Design of Intrusion Detection System Based on Artificial Neural Network and Application of Rough Set

Securing data in a networked environment has been a major concern for Network Administrator as intruders may get access and steal the information available in the Computer network. As absolute security is not possible in a network, detecting intrusion is very important from the standpoint of protection of the information as well as the network. The paper intends to cover the development of an Intrusion Detection System based on Neural Network Systems. As the Intrusion Detection Systems (IDS) have to depend on known signatures, we have to train the IDS about the signatures. KDD99 is a freely available dataset for intrusion signatures and we intend to use KDD99 dataset for both training and testing our IDS. As the number of input attributes for the signatures to the IDS (for detection of the intrusion of the network) is quite high(41 in all together), minimization of the inputs to the network is very important as processing time for the inputs to be kept minimized for real time detection. Selection of the most relevant features is very important for this and concept of Rough Set has been applied for selection of the most relevant features. Effects of minimization of input features for the signatures, through use of Rough Set for Detection of Intrusions in a network, have been studied in this research paper.