Covert UAV Data Transmission Via Semantic Communication: A DRL-Driven Joint Position and Power Optimization Method

The integration of covert unnamed aerial vehicle (UAV) data transmission and semantic communication has recently shown great potential to improve efficiency and reliability of data transmission. However, due to the variability of the complex communication environment, existing researches are difficult to provide reliable optimization method for UAV under adversarial eavesdropping scenarios. In this paper, we propose a covert UAV data transmission via semantic communication (CUDT-SC) framework, leveraging a full-duplex (FD) UAV to effectively obscure the entire transmission process from eavesdroppers. Furthermore, a newly-defined metric, namely semantic UAV data throughput (SUDT), is introduced to quantify the system’s performance. Building on this foundation, we propose a deep reinforcement learning driven joint position and power optimization (DRL-JPPO) algorithm to maximize the accumulated SUDT during the UAV data transmission period. Extensive experiments demonstrate the effectiveness of CUDT-SC framework. Specifically, the designated DRL-JPPO algorithm not only attains a significantly higher accumulative SUDT of up to $\mathbf{4 0 \%}$, but also demonstrates rapid convergence when compared to the benchmark schemes.