Intrusion Detection System using Improved Pelican Optimization Algorithm-based Support Vector Machine to Secure Data in Cloud
R. Radhika · 2024
Cloud technology encompasses servers accessible via the internet, along with the software and databases hosted on those servers. Cloud has become popular among users in numerous areas with maximized efficiency and reliability. However, security is a significant issue in the cloud due to the cloud service being accessed through the Internet by various users. In this research, the Improved Pelican Optimization Algorithm-Support Vector Machine (IPOA-SVM) is proposed for the Intrusion Detection System (IDS) to secure data in the cloud utilizing Machine Learning (ML). Initially, data is acquired from CIC-IDS2017 and UNSW-NB15 benchmark datasets and is used to evaluate the IPOA-SVM technique. The min-max normalization is established for normalizing obtained data. The Feature Selection (FS) is performed utilizing the IPOA approach to select the appropriate features. The POA approach is improved by using a Tent chaotic map for initializing pelicans and a dynamic weight factor for helping the pelican to update its position constantly. At last, SVM is employed to detect and categorize attacks effectively and accurately. An IPOA-SVM achieves better accuracy of 99.96% and 99.86% compared to existing techniques like Ensemble Learning (EL), two-phase IDS, Ensemble-based Automatic Feature Selection (EAFS), Information Gain-Chi-Square-Particle Swarm Optimization-Random Forest (IG-CS-PSO-RF), and Self-adaptive Evolutionary Extreme Learning Machine (SaE-ELM) respectively.