Intrusion detection system in cloud computing by utilizing VTR-HLSTM based on deep learning

Abstract

Cloud computing (CC) is a rapidly developing IT approach with intrusion detection system being a crucial tool for safeguarding virtual networks and machines from potential threats, thereby mitigating security concerns in the cloud environment. The intrusion detection system (IDS) system demands significant improvements, primarily based on optimizing performance and bolstering security measures. This research aims to implement an IDS in cloud computing utilizing deep learning (DL) method. The DL model is a promising technique and is widely used to detect intrusions. The implemented hierarchical long short-term memory (HLSTM) method’s performance is evaluated for feature selection through variance threshold-based regression (VTR) on two IDS network datasets: Bot-IoT and network security lab-knowledge discovery and data mining (NSL-KDD). This paper concludes the use of an intrusion detection network resulting in high security and performance. Moreover, the implemented method on the NSL-KDD and Bot-IoT datasets obtains respective accuracies of 99.50% and 0.995. It is compared with the existing methods namely, ensemble ID model for CC utilizing DL, LeNet, fuzzy deep neural network with a Honey Bader algorithm for privacy-preserving ID, and improved metaheuristics with a fuzzy logic-based IDS for cloud security, and beluga whale-tasmanian devil optimization based on deep convolutional neural network (CNN) with TL, chronological slap swarm algorithm-based deep belief network (DBN), and dragonfly improved invasive weed optimization-based Shepard CNN.