Deep Reinforcement Learning Based Intelligent Reflecting Surface Assisted Secure Hybrid Visible Light and Millimeter Wave Communications

Abstract

In the contemporary era of constant connectivity and increasing data demands, ensuring secure, efficient, and reliable wireless communications is crucial. This dissertation addresses the integration of millimeter waves (mmWaves) and Visible Light Communication (VLC) technologies, coupled with recent advancements in Physical Layer Security (PLS) domain. The research employs Intelligent Reflective Surfaces (IRS) to improve the confidentiality, performance, and reliability of the wireless network. Furthermore, the study incorporates the Deep Deterministic Policy Gradient (DDPG) method from Deep Reinforcement Learning (DRL) to optimize the system’s Secrecy Capacity (SC) comprehensively. The integration of IRS into the hybrid system adds a layer of dynamic security, enabling dynamic control of the communication environment and providing innovative ways to enhance security without depending on traditional security protocols. The DDPG method is introduced to determine the optimal SC of the system. This DRL approach enables the system to adapt its security protocols in real-time scenarios, responding to environmental dynamics and potential threats. Additionally, the DDPG algorithm can adapt to channel variations and high-dimensional factors. Furthermore, the algorithm will intelligently select the optimal technique to improve SC, whether VLC or mmWaves, while allowing efficient power management. Integrating mmWave and VLC technologies with advancements in PLS, IRS, and DDPG provides a robust strategy for optimizing the security of the proposed wireless transmissions. The results of this thesis, through a theoretical framework and comprehensive simulations, contribute to the development of secure, efficient, and resilient future wireless communication networks that are capable of meeting the demands of modern connectivity.