基于密度加权的K均值聚类密钥量化方案

在物理层密钥生成过程中，现有传统的量化方案基于规则量化边界设计，无法根据实际采样测量值自适应动态调整量化边界，当量化边界不规则时边界附近量化产生的密钥不一致率较高。针对上述问题，提出一种基于密度加权的K均值聚类密钥量化方案，通过统计测量值在复平面的分布疏密情况，使量化分界远离测量值密集区域，降低噪声对量化判决的影响仿真分析表明，该方案可实现自适应调整量化区域，使区域分界附近测量值分布最少。在Ibit量化、SNR=10AB条件下密钥不一致率可达0.03，与传统K均值量化方案相比，降低了生成密钥的不一致率。

Physical Layer Secret Key Quantization Scheme Based on Density-weighted K-means Clustering

In the physical layer key gexisting traditional quantizadesigned based on the regular quantization boundary, which can not adjust the quantizationboundary adaptively and dynamically according to the actual sampling measuremenWhen thequantization boundary is irregular, the key inconsistency rate near the boundary isabove problems, this paper proposes a key quantization scheme based on derclustering. By counting the density distribution of the measured values in the complex plane, thequantization boundary is far away from the dense area of the measured values, and the influence ofnoise on the quantization decision is reduced. Simulation analysis shows that the proposed schemecan adaptively adjust the quantization boundary to minimize the distribution of measured values nearthe boundary. Under the condition of I bit quantization and SNR= 10dB, the key inconsistency rateof this scheme can reach 0. 03. Compared with the traditional K-means quantization scheme, thisscheme has a sigmificant improvement in the key agreement rate.