Network Intrusion Detection System Using Honeypot in cloud

Abstract

The rapid growth in internet usage has led to a surge in cybercrime, data breaches, and site-hosting vulnerabilities, posing significant challenges to online security. Cloud computing has become a key solution, offering reliable, scalable, and cost-effective services, but it also introduces new security risks due to its open and interconnected nature. As cyber threats evolve, the need for advanced intrusion detection systems has become more pressing. Honeypots, decoy systems designed to lure attackers, have proven to be an effective tool for identifying malicious activity and diverting harmful traffic away from critical infrastructure. However, the effectiveness of honeypots in cloud environments can be enhanced by leveraging machine learning models for more accurate and efficient intrusion detection. This study investigates the integration of machine learning models, specifically Recurrent Neural Networks (RNN) and Convolutional Neural Networks (CNN), to improve the detection of intrusions in cloud-based honeypots. By analyzing large datasets of network traffic, the models were trained to distinguish between normal and malicious activity. The results demonstrate that the RNN model achieved an accuracy of 93.29%, while the CNN model reached 91.57%, highlighting their potential for high-accuracy intrusion detection. These findings underscore the significant role of deep learning in enhancing cybersecurity by enabling faster detection and response to cyber threats in cloud environments. The study also discusses the challenges and considerations of implementing honeypots in cloud infrastructures, such as privacy concerns and legal issues when dealing with third-party cloud service providers. The combination of honeypots with advanced machine learning techniques, including RNN and CNN, offers a promising direction for future research in cloud security.