Deep Learning-Assisted Secure UAV-Relaying Networks With Channel Uncertainties

Xiao Tang, Na Liu, Ruonan Zhang, Zhu Han

科研成果: 期刊稿件文章同行评审

14 引用 (Scopus)

摘要

Unmanned aerial vehicles (UAVs) play a critical role for 5 G/B5G communications. In this paper, we investigate the physical layer security issue for ground transmissions with the help of a UAV. Specifically, we consider the scenario without direct transmissions and employ the UAV as a mobile relay to amplify and forward the confidential information. Meanwhile, the receiver transmits cooperative jamming attempting to disrupt eavesdropping. We investigate the UAV deployment and receiver jamming strategy jointly for the optimized secrecy rate, while considering the channel uncertainties. In particular, for the case of complete channel state information (CSI), the jamming power is obtained by bi-section search, and we propose a data-trained deep neural network (DNN) to approximate the optimal UAV deployment. Meanwhile, for the case with partial CSI, we design a deep Q network (DQN) to learn the optimal UAV deployment, due to the increased complexity to determine the jamming strategy with channel statistics. Simulation results demonstrate that both the learned DQN and trained DNN model effectively approximate the optimal deployment of the UAV. Moreover, compared with the baselines, our proposal significantly improves the security performance of legitimate transmissions.

源语言英语
页(从-至)5048-5059
页数12
期刊IEEE Transactions on Vehicular Technology
71
5
DOI
出版状态已出版 - 1 5月 2022

指纹

探究 'Deep Learning-Assisted Secure UAV-Relaying Networks With Channel Uncertainties' 的科研主题。它们共同构成独一无二的指纹。

引用此