A 5G Physical Layer Key Generation Scheme Based on Attention-Integrated Autoencoder

The fifth generation (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{5 G}$</tex>) communication is progressively being commercialized and civilianized. It is crucial to protect the wireless transmission of sensitive information. Physical layer key generation (PLKG) emerges as a lightweight symmetric key generation technique based on channel reciprocity, which can further ensure the security of 5 G communications. Aiming at the problem of weak channel reciprocity in 5 G communications, this paper proposes a 5G PLKG scheme based on an attention-integrated autoencoder. Firstly, channel probing time slots are planned in accordance with the slot configuration characteristics of 5 G protocol to maximize the reciprocity of the extracted channel features on the time scale. Secondly, an attention-integrated autoencoder model is proposed to learn the common features in the channel measurements of Alice and Bob. Simulation results show that the proposed method performs better in terms of channel reciprocity enhancement.