Nazia Sarwat Islam
Prof. Dr.-Ing. Werner Henkel
Physical-layer security options can provide robust security measures in a wireless scenario and is a topic of current interest. In this work, we consider a wireless, reciprocal channel between two legitimate users and assume a frequent physical-layer key generation scheme to provide encryption security. The key generation scheme is based on a vector quantization method which yields almost identical keys on both ends of the channel, with differences arising from independent noise contributions at the two ends. Assuming one user’s key to be the correct one, we focus on key reconciliation at the other end of the link in order to obtain the same key. To this end, we employ a Slepian-Wolf key reconciliation scheme using LDPC codes (soft decoding). We require an irregular and multi-edge-type code for the reconciliation scheme. The decoder optimization is based on a rate maximization criterion and uses a linear program to deliver the optimized variable-node degree distributions, starting from a constant check-node degree-distribution polynomial. For the optimization, we apply density evolution in the convergence criterion. In this work, we have mainly focused on deriving the Log-Likelihood Ratios (LLRs) for the soft-decoder and subsequently designing the density evolution steps. We provide necessary derivation steps and also the results of the linear optimization program. A preliminary BER simulation is also presented.