Network Intrusion Classifier with Optimized Clustering Algorithm for the Efficient Classification

Nowadays network security is becoming more and more challenging due to increasing number of network attacks. The major attack which carried out on the network seems to be DoS attack, DDoS attack, Botnet Attack and Bruteforce Attack. The proposed research work will classify the incoming packets in the cloud environment (through network) as attack or non-attack packets with best clustering and cluster head optimization algorithm. The overall classified cluster lead to further classification on the specified four major attacks with the better accuracy and other network intrusion measures. There are sorne fixed techniques associated in this classification and those are Principal Component Analysis PCA for extracting the features and AutoFncoder (classifier algorithm based on deep-learning) for classifying the attacks after clustering as attack and non-attack group. This research includes a CSE--CIC--IDS dataset which had more than 1.8 crore packets. These are the packets which are having above four major attacks and also includes SQL Injection and Infiltration attacks in lesser proportionate. The motive behind this research is to find the best fit of clustering algorithm with a combination of efficient cluster head optimization technique with the fixed technique of PCA based feature reduction and deep learning based classifier on to and from of clustering mechanism for betterment of intrusion classification accuracy and other network measures.