| In recent years,with the continuous upgrade of wireless communication technology,especially with the commercialization of 5G technology,the application of 5G technology in daily life has become more and more extensive,and the integration and interaction of information has increased.Communication technology,especially wireless communication,is closely related to people's daily life and even the development of national society.The problem of wireless communication' security has attracted widespread attention,and the security methods of wireless communication have increasingly become the focus of attention.In the meanwhile,due to resource constraints,the technology of using the physical layer channel characteristics to generate keys to achieve secure communication has become one of the important methods to ensure the security of wireless communications.Based on a typical passive eavesdropping model,this paper conducts research on the key generation ways using the characteristics of the wireless channel of the physical layer.Aiming at the inconsistency between the key consistency and the generation rate in the key generation process caused by the incomplete utilization of channel information in the feature quantization process,combined with the clustering algorithm in machine learning,an algorithm for channel feature quantization using K-Means clustering is designed to optimize the key generation method based on channel features.On the basis of ensuring the speed of key generation,the matching of the initial key generation is improved to reduce the interaction between the two communication parties in the subsequent steps of key generation and realize confidential communication.Aiming at the problem that the large amount of feature quantization data in key generation causes a large amount of calculation and the choice of the initial clustering center in the K-Means clustering algorithm will cause different clustering effects to affect the key consistency.The key generation method has been improved as follows:(1)it is proposed to use PCA to pre-process the channel features before quantization to achieve the purpose of reducing the dimension,thereby reducing the amount of data.And based on the characteristics of the cluster center distribution in K-Means clustering quantification,it is improved to achieve the purpose of fast clustering,thereby increasing the key generation rate.(2)Apply the binary K-means clustering algorithm to the quantization of key generation features to optimize the key generation method based on channel characteristics and improve the consistency of the initial key. |
