A Comparative Evaluation of Machine Learning-Based Intrusion Detection Systems for Securing Cloud Environments

Abstract

Cloud computing has advanced significantly alongside the growth of communication technology and data exchange. Many businesses and organizations now adopt cloud computing solutions and services to enhance flexibility and scalability. However, despite its numerous advantages, cloud computing remains increasingly susceptible to various security threats that can disrupt services and business operations. This highlights the critical need to strengthen the security of cloud environments. In this context, implementing robust protection measures, such as Intrusion Detection Systems (IDS), is essential to mitigate potential threats and safeguard sensitive data. To effectively counter the ever-evolving cyber threats landscape, IDS must possess adaptive capabilities. Hence, integrating Machine Learning (ML) technologies is imperative for the detection of a broad and diverse range of cyber threats, thereby enhancing the overall bolstering the security posture of the environment.
This research explores the integration of ML technologies in IDS and examines the application of feature selection methods to identify the key and most significant indicators for attack detection. The study conducts a comparative analysis of five ML techniques, employing two distinct feature selection methods to evaluate their effectiveness in strengthening the security of cloud environments. Using a recently developed, reputable dataset and concentrating on attack types that pose significant threats to cloud environments, our experimental results offer a comprehensive evaluation of these techniques, including a variety of machine learning algorithm performance metrics.