A Hybrid Intrusion Detection System with a New Approach to Protect the Cybersecurity of Cloud Computing

TL;DR

This paper introduces a novel hybrid intrusion detection system (HyIDS) that uses an Energy-Valley Optimizer for feature selection and machine learning for threat detection, significantly improving accuracy and detection rates in cloud cybersecurity.

Contribution

The main contribution is the development of a new hybrid intrusion detection approach utilizing EVO for feature selection, outperforming existing optimizers like GWO.

Findings

EVO achieves higher accuracy than GWO in feature selection.

The proposed HyIDS attains over 99% accuracy on multiple datasets.

EVO improves overall system performance in threat detection.

Abstract

Cybersecurity is one of the foremost challenges facing the world of cloud computing. Recently, the widespread adoption of smart devices in cloud computing environments that provide Internet-based services has become prevalent. Therefore, it is essential to consider the security threats in these environments. The use of intrusion detection systems can mitigate the vulnerabilities of these systems. Furthermore, hybrid intrusion detection systems can provide better protection compared to conventional intrusion detection systems. These systems manage issues related to complexity, dimensionality, and performance. This research aims to propose a Hybrid Intrusion Detection System (HyIDS) that identifies and mitigates initial threats. The main innovation of this research is the introduction of a new method for hybrid intrusion detection systems (HyIDS). For this purpose, an Energy-Valley Optimizer (EVO) is used to select an optimal feature set, which is then classified using supervised machine learning models. The proposed approach is evaluated using the CIC_DDoS2019, CSE_CIC_DDoS2018, and NSL-KDD datasets. For evaluation and testing, the proposed system has been run for a total of 32 times. The results of the proposed approach are compared with the Grey Wolf Optimizer (GWO). With the CIC_DDoS2019 dataset, the D_TreeEVO model achieves an accuracy of 99.13% and a detection rate of 98.941%. Furthermore, this result reaches 99.78% for the CSE_CIC_DDoS2018 dataset. In comparison to NSL-KDD, it has an accuracy of 99.50% and a detection rate (DT) of 99.48%. For feature selection, EVO outperforms GWO. The results of this research indicate that EVO yields better results as an optimizer for HyIDS performance.