Neural-network-enabled Optical Steganography Based on Bias Tuning of Electro-optical Modulator
Weidong Shao, Mengfan Cheng, Lei Deng, Qi Yang, Xiaoxiao Dai, Deming Liu · 2022
We propose an optical steganography strategy based on precise bias tuning of an electro-optical modulator and a Convolutional Neural Network (CNN)-enabled extractor. Underneath the 10 Gbit/s public data transmission, the 195 Mbit/s stealth data is modulated on the bias states of the modulator based on the precise and stable bias control technique. The stealth embedment has been verified to make almost no impact on the public channel within a 20-km distance. For a legal stealth receiver, the “invisible” stealth data can be extracted with the aid of the CNN-enabled extractor. Only one optical source is needed in the proposed optical steganography scheme and the stealth channel is integrated into the public channel closely and naturally. The privacy and security of stealth data are analyzed.