Abstract:
Information security remains a paramount concern in wireless networks. Covert communication
offers a unique solution by hiding the transmission, making it undetectable
by the warden (Willie). This thesis investigates the approach for maximizing covert
data rate in a scenario where a multi-antenna transmitter (Alice) transmits to a covert
user (Bob) and a public user (Roy) through a UAV-IRS assisted relay, leveraging Non
Orthogonal Multiple Access (NOMA). The public user served by NOMA acts as artificial
noise, masking the covert transmission from a warden (Willie). The system
leverages Unmanned Aerial Vehicles (UAVs) for flexible deployment and Intelligent
Reflecting Surfaces (IRSs) to manipulate signal propagation. Deep Deterministic Policy
Gradient-A Deep Reinforcement Learning (DRL) approach is employed to optimize
both transmit beamforming at the transmitter and phase shifts at UAV-IRS, maximizing
the covert data rate at Bob, ensuring at least a minimum data rate at Roy while
minimizing signal strength at the warden. This UAV-IRS assisted NOMA with the
DRL approach offers a promising solution for secure and reliable covert communication
in various scenarios.
