A Secure Deep Joint Source-Channel Coding Method for UAV-Assisted Image Transmission
作者：蒋雨航 时间：2026年10月01日 13:49 点击数：
语言：English
发表会议：2026 IEEE/CIC International Conference on Communications in China (ICCC)
发表日期：2026年8月9日
摘要：
With the evolution of sixth-generation (6G) networks, the integration of semantic communication and unmanned aerial vehicle (UAV) communication holds great promise for further enhancing the efficiency and reliability of wireless networks. Nevertheless, due to the openness of wireless communication channels, UAV-assisted semantic transmission remains vulnerable to security threats such as interference and eavesdropping. To address the rapidly varying channel characteristics of UAVs while effectively countering eavesdropping, this paper proposes a secure deep joint source-channel coding (S-DJSCC) method, which jointly models and optimizes physical-layer beamforming and semantic channel coding in an end-to-end manner, enabling the simultaneous enhancement of communication performance and system security. By synergizing semantic security with physical layer security (PLS), S-DJSCC bolsters the secure transmission capabilities for legitimate users across diverse scenarios. Furthermore, to objectively evaluate the security performance, we define a novel metric termed joint semantic secrecy rate (JSSR), derived from the transmission secrecy rate. Simulation results show that the proposed method significantly improves the JSSR by up to 82.2% over conventional PLS methods. Notably, in extreme scenarios with overlapping directions, conventional PLS methods fail entirely, while the proposed method only degrades by at most 5.7%.
原文链接
