Quantum Deep Reinforcement Learning for Secure Multi-User UAV RSMA Networks

In this work, we investigate the enhancement of secrecy rate maximization in multi-user unmanned aerial vehicle (UAV) rate-splitting multiple access (RSMA) networks under stringent secrecy constraints. The objective is to maximize the aggregate secrecy rates of all users, ensuring robust secure communication. To address this, the original problem is decoupled into two interconnected subproblems: joint beamforming and rate allocation, and UAV trajectory optimization. For the beamforming and rate allocation subproblem, we propose a novel quantum deep reinforcement learning (QDRL) approach that leverages quantum computing advantages to enhance learning efficiency and solution quality. The QDRL framework utilizes quantum neural networks to process complex channel state information and make optimal decisions in high-dimensional spaces. For the UAV trajectory optimization, which is inherently nonconvex, we employ quantum-enhanced policy gradient methods that significantly improve convergence speed and solution robustness compared to classical approaches. Simulation results validate the effectiveness and robustness of the proposed quantumbased framework in enhancing secrecy rates while ensuring secure communication, demonstrating superior performance over conventional optimization techniques.