Stochastic Perturbation-based Physical-Layer Secret Key Generation in Quasi-Static Scene

A secret key generation scheme based on stochastic perturbation is proposed to improve the generated secret key rate for quasi-static channels in Internet of Things (IoT). First, the two legitimate nodes send training sequences to estimate the channel information and the initial key is obtained by quantization. Second, each sub-key in the initial key is classified, which the classification label is the column value in the stochastic perturbation matrix. Finally, each sub-key is combined with the corresponding stochastic perturbation sequence through XOR operation to obtain the perturbed key estimation sequence. The simulation results show that with the same SNR, the key generation rate based on stochastic perturbation is faster, and the correlation between the sub-keys is lower. The National Institute of Standards and Technology (NIST) randomness test shows that keys generated based on stochastic perturbation have better randomness.