Intrusion Detection System: An Ensemble Deep Learning Approach for Cloud Computing Using EBWO

Cloud computing is the industry standard for data storage, sharing, processing, and other services. It experienced numerous security problems as a result of the regular attacks. These security issues are worsened by the variety of attack situations that exist. One of the most established safety measures applied to cloud computing is the intrusion detection system (IDS). An effective security model is necessary for the IDS system, though, to increase cloud security. In this study, we used ensemble categorization methods and a feature selection algorithm to construct an effective IDS for the cloud environment. The proposed BOT-IOT, CSE-CIC-IDS 2018, and Ciciddos datasets were pre-processed, which involved cleaning the data, applying one hot encoding, and normalizing steps. The Enhanced Black Widow Optimization (EBWO) algorithm was employed to choose the most advantageous reduced feature sets from the provided incursion datasets. We used an ensemble of Hierarchical Multi-scale LSTM (HMLSTM) and Darknet Convolutional Neural Network (DNetCNN) to categorize the attacks. The combination of DNetCNN and HMLSTM was used to identify intrusions, effectively classifying attacks, lowering false alarm rates, and increasing detection rates. Simulation research showed that the proposed strategy performed better than the baseline in terms of F-Score, DR, and FPR, as well as accuracy, detection rate, and precision.