An intelligent intrusion detection and prevention system using hybrid deep learning techniques in cloud environment

Abstract

In the current digital era, cloud computing has become an essential newlinetechnology that provides on-demand network access to a shared pool of newlineconfigurable computing resources. However, ensuring the security and newlineprivacy of cloud environments remains a significant challenge due to their newlineopen and distributed architecture, which makes them vulnerable to potential newlineintruders. To address this concern, hybrid deep learning techniques have been newlineproposed in this research study to develop an intelligent intrusion detection newlineand prevention system for cloud environments. This study intelligent intrusion newlinedetection and prevention system using hybrid deep learning techniques in a newlinecloud environment focuses on the major objectives of detecting and newlineclassifying malicious packets, enhancing detection accuracy, monitoring and newlinepreventing malicious packets in the cloud environment. newlineAn intelligent intrusion detection and prevention system consists of newlineDeep Belief Network (DBN)-based Particle Swarm Optimization with Long newlineShort Term Memory (PSO-LSTM) and Sparse Auto-encoder followed by a newlineStacked Contractive Auto-encoder (S-SCAE)-based Bi-Directional LSTM newlinewith Dropout and Attention Layer (Bi-DLSTM-DAL) detection models. Each newlinedetection model encompasses a data collection and preprocessing subsystem, newlinea feature extraction subsystem, a detection subsystem along with a prevention newlinesubsystem. All of these subsystems utilize hybrid deep learning techniques newlineaimed at providing effective intrusion detection and prevention capabilities newlinefor cloud environments. To evaluate the system, the NSL-KDD dataset is newlineused, which comprises 41 features. Individual attacks are categorized as DoS, newlineProbe, R2L, U2R, as well as a normal class newline