Enhancing Secrecy in UAV RSMA Networks: Deep Unfolding Meets Deep Reinforcement Learning

TL;DR

The paper tackles secrecy-rate maximization in a downlink of $U$ UAVs serving $K$ legitimate users in the presence of $I$ eavesdroppers, formulating a nonconvex joint optimization over beamformers ${\bf w}_c, {\bf w}_{u,k}$, secrecy-rate allocation ${\mathfrak r}_{c,u,k}^{\sec}$, and UAV trajectories ${\bf q}_u(t)$. A novel multiagent DRL framework, DUN-DRL, combines deep unfolding to port an iterative beamforming-rate algorithm into a multi-layer network (HNet) with a CNN-based trajectory network, and employs deep deterministic policy gradient (DDPG) for learning. By casting the problem as a Markov decision process with state/action/reward definitions, the method achieves higher secrecy-rate performance than baselines, including near-optimal configurations, and demonstrates faster convergence. The approach offers a practical path to secure UAV-assisted RSMA deployments by merging model-driven unfolding with data-driven trajectory control.

Abstract

In this paper, we consider the maximization of the secrecy rate in multiple unmanned aerial vehicles (UAV) rate-splitting multiple access (RSMA) network. A joint beamforming, rate allocation, and UAV trajectory optimization problem is formulated which is nonconvex. Hence, the problem is transformed into a Markov decision problem and a novel multiagent deep reinforcement learning (DRL) framework is designed. The proposed framework (named DUN-DRL) combines deep unfolding to design beamforming and rate allocation, data-driven to design the UAV trajectory, and deep deterministic policy gradient (DDPG) for the learning procedure. The proposed DUN-DRL have shown great performance and outperformed other DRL-based methods in the literature.

Figures (5)

• Figure 1: Structure of the proposed DUN-DRL
• Figure 2: Detailed structure of HNet
• Figure 3: Achievable secrecy rate versus training episodes.
• Figure 4: Cumulative distribution functions (CDF) of the secrecy rate.
• Figure 5: Achievable secrecy rate versus UAV power.