Joint Design of Balance Mechanism and Compensation Strategy in K-means Clustering Assisted Physical Layer Key Generation

Clustering is an unsupervised machine learning (ML) algorithm, which is able to make the samples in the same cluster as similar as possible. Due to this characteristic, it is great for quantization in the physical layer key generation (PLKG). However, there are two problems that need to be solved urgently, one is the low entropy problem and the other is the cell boundary problem. In this paper, we propose a balance mechanism with updated distance difference to achieve uniform classification of channel samples. Meanwhile, to solve the cell-boundary problem, we propose a limited compensation strategy that reduces bit disagreement ratio (BDR) while leaving Eve with limited access to information. The simulation results verify that our proposed scheme effectively solves the above problems and the BDR of our scheme is lower than that of the existing scheme.