Prof. Dr.-Ing. Werner Henkel
With increasing significance of wireless transmission, physical layer security becomes a challenging topic resulting from the Shannon’s limit. This work explores physical layer security from the aspect of information theory and discusses key generation and reconciliation by Slepian-Wolf coding
and low-density parity-check (LDPC) code. In order to design the LDPC code, we use density evolution on the log-likelihood ratio (LLR) distribution which is forwarded along the edges in Belief-Propagation (BP) algorithm. This paper focuses on the density evolution procedure when the measured distribution is non-consistent and non-Gaussian and follows the iterations step by step. Eventually, the research finds that though the initial measurement in Bob’s side is obviously non-Gaussian and non-consistent, after several iterations or even only convolutions with the densities
from other nodes within an iteration, a predictive result occurs. Though the final result in this work is still not Gaussian but the finding could say that the distribution tends to be consistent Gaussian. This result will contribute to the further studies of physical layer security.