Performance Comparison Analysis of Classification Methodologies for Effective Detection of Intrusions

Abstract

Intrusion detection systems (IDS) are critical in many applications, including cloud environments. The intrusion poses a security threat and extracts privacy data and information from the cloud. The user has an Internet function that allows him to store personal information in the cloud environment. The cloud can be affected by various issues such as data loss, data breaches, lower security and lack of privacy due to some intruders. A single intrusion incident can result in data within computer and network systems being quickly stolen or deleted. Additionally, intrusions can cause damage to system hardware, resulting in significant financial losses and exposing critical IT infrastructure to risk. To overcome these issues, the study employs the performance comparison analysis of Autoencoder Convolutional neural network (AE+CNN), Random K-means clustering assisted deep neural network (RF+K-means+DNN), Autoencoder K-means clustering assisted long short term memory (AE+K-means+LSTM), Alexnet+Bi-GRU, AE+Alexnet+Bi-GRU and Wild horse AlexNet assisted Bi-directional Gated Recurrent Unit (WABi-GRU) models to choose the best methodology for effective detection of intrusions. The data needed for the analysis is collected from CICIDS2018, UNSW-NB15 and NSL-KDD datasets. The collected data are pre-processed using data normalization and data cleaning. Finally, through this research, the best model suitable for effective intrusion detection can be identified and used for further processes. The proposed models, such as RF+K-means+DNN, AE+K-Means+LSTM, AlexNet Bi-GRU, AE+Alexnet+Bi-GRU and WABi-GRU can obtain an accuracy of 99.278%, 99.33%, 99.45%, 99.50%, 99.65% for the CICIDS dataset 2018 for binary classification. In multi-class classification, the AlexNet Bi-GRU, AE+Alexnet+Bi-GRU and WABi-GRU can attain accuracy of 99.819%, 99.852% and 99.890%. In NSL-KDD, the AlexNet Bi-GRU, AE+Alexnet+Bi-GRU and WABi-GRU achieve accuracy of 99.34%, 99.546% and 99.7%. In UNSW-NB 15 dataset, AlexNet Bi-GRU, AE+Alexnet+Bi-GRU and WABi-GRU achieve accuracy of 99.313%, 99.399% and 99.53%. AlexNet Bi-GRU-based models can obtain better performances than other existing models.