A Comparative Study for the Detection of Attacks on Cloud Computing Systems

Abstract

Cloud computing is a rapidly growing technology that utilizes internet-based computing power. Data, information, and other resources are provided to users on demand through computers and other devices. Cloud, a popular technology with flexible, cost-effective and configurable computing resources, provides computing resources on demand. Due to its flexible infrastructure, network-centric approach, and easy accessibility, cloud computing is increasingly being used by both small and large organizations. Despite all the benefits of cloud computing, there are concerns about how security can be ensured and how application security is maintained in a cloud environment, as this technology represents a new computing model. Innovations that come with cloud computing, such as multiple users sharing the same physical or virtual resources and organizations obtaining services from a provider, also bring some security risks. The uncertainty surrounding security hinders the progress of the cloud computing market. In this study, two different intrusion detection models were compared using deep learning techniques for cloud security. The two proposed models were trained and tested using the UNSWNB15 dataset. Experimental results show that the first model, based on CNN, has a lower accuracy rate compared to the second hierarchical model created using both CNN and RNN. Our experiments on the network traffic dataset demonstrate that utilizing both CNN and RNN in an intrusion detection system can enhance the ability to detect a wide range of potential security threats by leveraging the strengths of both models to identify patterns in network traffic data. © 2023 IEEE.