Securing the Metaverse: A Deep Reinforcement Learning and Generative Adversarial Network Approach to Intrusion Detection
Tarek Ali, Mohammed Al-Khalidi, Rabab Al-Zaidi, Amna Eleyan, Muhammad Atif Ur Rehman · 2024
This paper explores the pivotal domain of security within the Metaverse and proposes an innovative method to tackle the distinctive challenges it presents. The proposed solution leverages Generative Adversarial Networks (GANs) to generate synthetic data, and Deep Reinforcement Learning (DRL) as a targeted model. The proposed Intrusion Detection System (IDS) effectively navigates the intricate Metaverse environment. Incorporating GANs guarantees the production of diverse synthetic data, thereby mitigating concerns linked to class imbalance. Moreover, DRL empowers the IDS to differentiate between customary and unusual user actions. According to our research findings, our approach surpasses competitors, particularly when faced with synthetic or augmented data, considering virtual space, user interactions, and network activities. This serves as evidence that our model possesses the capability to enhance intrusion detection in the dynamic Metaverse environment.