Novel intelligent blind information reconciliation for LDPC codes in quantum key distribution systems

Abstract

Information reconciliation is an important step in quantum key distribution (QKD). As a good channel coding technology, LDPC (low-density parity-check) code perform well in information reconciliation. In this paper, we investigate a new approach, namely blind information reconciliation. In our approach, after the state distribution is carried out by Alice and the measurements are taken by Bob, Alice can select an appropriate LDPC encoder (parity-check matrix) from the predefined candidate pool according to the estimated QBER (quantum bit-error-rate). Bob can use the expectation–maximization (EM) estimator and the maximum average log-likelihood ratio (LLR) detector to blindly identify parity-check matrices associated with not only different code-rates but also different codeword-lengths Alice employed. Moreover, puncturing and shortening can also be invoked to adjust the code-rate slightly to improve the error-correction performance. Monte Carlo simulations have demonstrated that our proposed new blind information-reconciliation scheme can successfully identify the parity-check matrix even under high QBER conditions, reduce the interactive communication-rounds, and improve the reconciliation efficiency in comparison with the conventional rate-adaptive LDPC coding scheme without QBER estimation.