An Optimal Cluster Based Intrusion Detection System for Defence Against Attack in Web and Cloud Computing Environments

In recent years, the rapid growth of information technology organizations causes ability to meet their demands such as scalability, mobility and flexibility. The security and privacy is a major issue of those organizations that’s why they move the data to the cloud. Meanwhile, the security in cloud has become an important issue in the growing demand of cloud computing, Due to the nature characteristics of cloud, those confidential data are vulnerable to attacks/malicious or intruders. Several Intrusion Detection System (IDS) have been proposed for cloud environment to enhance the security problem but they are not possible to solve those issues with better accuracy, due to the recent real-time intruders. However, those IDSs are possible to solve and resist limited and known attacks. In this paper, we propose Optimal Cluster based Intrusion Detection System for defence against attacks in web and cloud computing environments (OC-IDS). We use hybrid optimization algorithm i.e. Multi-verse is combined with Chaotic Atom search optimization (MCA) algorithm for pre-processing which removes the unwanted/repeated data in dataset. We introduce a Chaotic Manta-ray Foraging Optimization (CMFO) based clustering technique which segment the data in different groups. Then, we develop hybrid machine learning technique i.e. Modified Teacher Learning based Deep Neural Network (MTL-DNN) which categorize the attack in cloud environment as a novelty of this study. Finally, the proposed OC-IDS technique can evaluate through standard open source datasets are KDD cup’99 and MSL-KDD, the performance of proposed and existing techniques are compared with different metrics such as accuracy, precision, recall and F-measure. Our proposed OC IDS MTL-DNN attains 95.01% accuracy in KDD cup’99 dataset.