Physical Layer Key Generation Scheme for MIMO System Based on Feature Fusion Autoencoder
Yanru Chen,Zhengyu Chen,Yuanyuan Zhang,Zhiyuan Luo,Yang Li,Bin Xing,Bing Guo,Liangyin Chen
DOI: https://doi.org/10.1109/jiot.2023.3288641
IF: 10.6
2023-01-01
IEEE Internet of Things Journal
Abstract:Recently, the use of wireless channel state information (CSI) to generate encryption keys in the physical layer has gained significant attention from researchers. Unlike classical cryptography, this approach relies on the variability of the wireless channel, channel reciprocity, and spatial decorrelation to ensure security, making it more lightweight and providing strong randomness. This article proposes a physical layer key generation scheme for wireless LAN MIMO systems based on feature fusion autoencoder (FFAEncoder) to address the issue of a high key disagreement rate (KDR). Our approach involves extracting amplitude and phase features separately, fusing them through multiplication operator in a neural network, and using an autoencoder to extract common features. The proposed scheme was evaluated on multiple data sets in different real-world scenarios, and it was found that the transmitter and receiver codeword’s mean squared error (MSE) and mean absolute error (MAE) were smaller than those of the current models, indicating better key generation performance. Additionally, the proposed scheme’s KDR was smaller, with a decay rate faster than that of the other two models in the same environment, and the primary key bits were 1/2 and 1/3 that of the other models, respectively, as the signal-to-noise ratio (SNR) increased.
computer science, information systems,telecommunications,engineering, electrical & electronic