Chronological salp swarm algorithm based deep belief network for intrusion detection in cloud using fuzzy entropy

Abstract

Cloud computing is susceptible to the existing information technology attacks, as it extends and uses the traditional operating system, information technology infrastructure, and applications. However, in addition to the existing threats, the cloud computing environment faces various security issues in detecting anomalous network behaviors. In order to resolve the security issues, an effective intrusion detection system named Chronological Salp Swarm Algorithm‐based Deep Belief Network is proposed for detecting the suspicious intrusions in a cloud environment. Accordingly, the proposed Chronological Salp Swarm Algorithm‐based Deep Belief Network is developed by integrating the Chronological concept with the Salp Swarm Algorithm. The optimal solution for detecting the intrusion is revealed using the fitness function, which accepts the minimal error vale as the optimum solution. Here, the weights are optimally tuned by the proposed algorithm to generate an effective and optimal solution for detecting the intruders. The proposed Chronological Salp Swarm Algorithm‐based Deep Belief Network obtained better performance through the facility of exploitation and the exploration in search space. The performance of the proposed method is analyzed using two datasets, namely KDD cup dataset and BoT‐IoT dataset, the comparative analysis is performed with the existing methods, such as Host based intrusion detection system, Deep learning, and Deep Neural Network + Genetic Algorithm. The proposed Chronological Salp Swarm Algorithm‐based Deep Belief Network obtained better performance in terms of accuracy, sensitivity, and specificity, with the values of 0.9618%, 0.9702%, and 0.9307% using KDD cup dataset, and 0.9764%, 0.9824%, and 0.9309% using BoT‐IoT dataset.

Intrusion detection framework for securing privacy attack in cloud computing environment using DCCGAN‐RFOA

“…accuracy, specificity, sensitivity, F‐score, precision, false positive rate, AUC is analyzed. Then the proposed IDS‐CC‐CCGAN‐WSOA method is examined and likened with the existing methods, such as intrusion detection framework on cloud computing with deep belief network with salp swarm algorithm (IDS‐CC‐DBN‐CSSA), 31 intrusion detection framework on cloud computing using deep neural network with improved genetic algorithm and simulated annealing algorithm (IDS‐CC‐DNN‐IGASAA), 32 intrusion detection for cloud computing with multilayer perceptron neural network and artificial bee colony optimization algorithm (IDS‐CC‐MLPNN‐ABC), 33 intrusion detection for cloud computing using recurrent convolutional neural network with ant lion optimization algorithm (IDS‐CC‐RCNN‐ALO), 34 hybrid intrusion detection using map reduce based black widow optimized convolutional long short‐term memory neural networks (IDS‐CC‐BWO‐CONV‐LSTM), 35 and unified deep learning approach for efficient intrusion detection system using integrated spatial‐temporal features (IDS‐CC‐OCNN‐HMLSTM), 36 respectively. The simulation parameter of IDS‐CC‐DCCGAN‐RFOA is given in Table 3.…”

“…Some of the recent works are reviewed here. Karuppusamy et al 31 have presented an IDS-CC-DBN-CSSA. The aim was determine the security issues and identify the distrustful intrusions in cloud environment.…”

“…Here, feature selection process is done with the help of univariate ensemble feature selection method 21 The selected features are given to the DCCGAN 29 for classification which classifies the data as secured data and privacy attack data. Mainly, DCCGAN classifier does not illustrate some adopting optimization methods for computing the optimum variables for ensuring precise detection of intruder in CC. In this article, the red fox optimization algorithm 30 is proposed for optimizing the DCCGAN. Here, the proposed technique is implemented in MATLAB, its performance is estimated with performance metrics, like accuracy, sensitivity, specificity, precision, F score, and recall. The comparison of evaluation metrics is analyzed with the various existing methods, like IDS‐CC‐DBN‐CSSA, 31 IDS‐CC‐DNN‐IGASAA, 32 IDS‐CC‐MLPNN‐ABC, 33 IDS‐CC‐RCNN‐ALO, 34 IDS‐CC‐BWO‐CONV‐LSTM, 35 and IDS‐CC‐OCNN‐HMLSTM, respectively 36 …”

“…• The comparison of evaluation metrics is analyzed with the various existing methods, like IDS-CC-DBN-CSSA, 31 IDS-CC-DNN-IGASAA, 32 IDS-CC-MLPNN-ABC, 33 IDS-CC-RCNN-ALO, 34 IDS-CC-BWO-CONV-LSTM, 35 and IDS-CC-OCNN-HMLSTM, respectively. 36 The rest of this article is organized as: Section 2 reviews the literature survey, Section 3 explains the proposed methodology, Section 4 demonstrates the results and discussion, and Section 5 concludes the article.…”

Cycle‐consistentgenerative adversarial network optimized with water strider optimization algorithm fostered intrusion detection framework for securing cloud computing environment

“…The preprocessing output is fed to feature selection. Here, extracted the optimum features depends on correlation feature selection 29 approach. The optimum features based, the data is categorized as normal and anomalous data with the help of cycle‐consistent generative adversarial network. Mainly, cycle‐consistent generative adversarial network classifier not express the adaptation of optimization strategies to scale the optimal variables and assure exact intruder detection in cloud computing. So, the water strider optimization approach is proposed to optimize the cycle‐consistent generative adversarial network classifier. The proposed approach is simulated in MATLAB using the benchmark dataset of NSL‐KDD. The performance metrics, such as accuracy, F‐score, specificity, precision, sensitivity, false positive rate, and AUC are evaluated. Then, the obtained results are compared with existing approaches, like chronological salp swarm algorithm‐base deep belief network for IDS in cloud computing (IDS‐CC‐DBN‐CSSA), 30 deep neural network based IDS utilizing hybrid optimization of improved genetic and simulated annealing algorithms for intrusion detection in CC (IDS‐CC‐DNN‐IGASAA) 31 and multiple layer perceptron neural network with artificial bee colony (IDS‐CC‐MLPNN‐ABC) 32 methods, respectively.…”