.png){kind=icon}
.png){kind=icon}
Return
Deep Learning for Covert Communication
.png){kind=icon}
PRE
AI
DOI：10.23919/JCC.fa.2023-0710.202409.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In recent years, deep learning has been gradually used in communication physical layer receivers and has achieved excellent performance. In this paper, we employ deep learning to establish covert communication systems, enabling the transmission of signals through high-power signals present in the prevailing environment while maintaining covertness, and propose a convolutional neural network (CNN) based model for covert communication receivers, namely DeepCCR. This model leverages CNN to execute the signal separation and recovery tasks commonly performed by traditional receivers. It enables the direct recovery of covert information from the received signal. The simulation results show that the proposed DeepCCR exhibits significant advantages in bit error rate (BER) compared to traditional receivers in the face of noise and multipath fading. We verify the covert performance of the covert method proposed in this paper using the maximum-minimum eigenvalue ratio-based method and the frequency domain entropy-based method. The results indicate that this method has excellent covert performance. We also evaluate the mutual influence between covert signals and opportunity signals, indicating that using opportunity signals as cover can cause certain performance losses to covert signals. When the interference-to- signal power ratio (ISR) is large, the impact of covert signals on opportunity signals is minimal.
Keywords:
convolutional neural network
covert communication
deep learning
Journal
IF:
3.1
Papers:
1.9K
Citations:
5.0K